I learned how to weld (MIG) and built a giant mushroom to house a mannequin I dubbed "the mushroom man" over about 100 hours in the last 4 weeks. I covered the outside with thick foam panels cut to size, cementing them in place with copious amounts of spray foam. I shaved the outside to a nice shape with a sawzall and the inside I covered in chicken-fenced, then attached a painters tarp to that (so it could be painted on).
To fit on a trailer (the mushroom's cap is 11.5ft wide) the cap comes off the stem and the edges of the cap are two half-moons which have fixed mounting points where threaded rod sticks through some welded washers, and a nut is put on in place. I was too last minute to install the 200 WS2811 pixels and have them run some cool patterns, before the music festival I brought it to came time, but even just a lantern on top (another painters tarp covered the cap's metal-frame, and everything was spray painted) looked great.
Super fun project. Expensive, but I learned a lot, got to be creative, and I'm happy to try out new things and make the best of my before-children time. Also, it was such a joy seeing people croud around the mushroom (and site beside the mushroom man inside) at night during the festival.
Are you an IRC shitposter? Isn't it hard to experiment with Toilet/Figlet fonts and flags? Well _not anymore._ Presenting Tuilet: a front-end to Toilet written by us, for us.
Built a site that will follow development of popular upstream open source projects on a daily basis and uses AI to summarize the commits and attempt to make it easier to track what's going on with the project: https://gitpulse.org
I’m working on https://www.zonehero.io/ to deliver a cost effective alternative to AWS native load balancing (ALB).
We already had a PoC which saved our customer 50% vs ALB (in their case that’s more than a million dollars a year), we’re now working on tooling and scaling the solution up!
Next we want to bring wasm modules to the lbs, to do edge traffic curation, bring your own model, etc.
It's essentially a simpler, read-only, AWS dashboard where everything is a filterable, searchable, exportable-to-CSV table, with some extra features like multi-region mode, saved notes, and a debugger for access denied errors.
It uses the AWS SDK for JavaScript, so everything is run client-side from your browser. I'm not 100% sure what direction I'm taking it yet, but it's been fun to hack on!
Thankfully programmatic. It’s a common UI table widget, essentially, and I’ve written some custom code to handle multi-region support, updating the AWS credential handler, pagination, and response processing. From there, it’s a matter of plugging in some common options for each AWS service: the service name, SDK method to call, pagination property (annoyingly, AWS API has numerous ways of paginating responses), etc. Takes about five minutes to add a new service.
I'm working on a 2D computational geometry library with a focus on irregular part packing in Rust. IE laying out laser cuts for wood or water jet for metal. https://crates.io/crates/babushka
I've got some nice types set up and a no fit polygon algorithm. Working on the genetic algorithm for packing à là svgnest.
Basically, my rhythm-based roguelike on original NES now has a proper economy, with gold gain and shops to spend the gold in. It also now supports PAL and Dendy systems, which is especially wonky due to the different framerate, but helped a bit by this being a rhythm game. As long as the music plays at the correct tempo, the rest of the game adapts its speed and "feels" correct at the lower framerate.
Tons of work left to do, most of it pixel art (I'm learning as I go) but it's progressing quite nicely.
I'm building a greenhouse. The frame is done, and I've got plastic and a door on it. Next, I'd like to build boxes to hold soil and allow for easy watering.
I'm working on https://nuenki.app/, a language learning tool. It teaches you a language while you procrastinate by inserting translations of appropriate-difficulty sentences into webpages as you browse HN etc.
Currently trying to reduce costs by switching from using DeepL (high quality, low latency, high cost) everywhere to a hybrid that also uses Claude (high quality, high latency, low cost) for text that is far from the user. Also experimenting with Gemma 2 9B via Groq to go in between them, but it's bad at following instructions and I don't quite trust the quality numbers I'm seeing for it (they're benchmarked with gpt-4o as a judge).
I'm also trying to work out marketing. I'm not good at it, and I dislike it, but I need to get good at it. Currently considering Reddit ads for awareness, some content marketing going over the technical details (there's some fun language processing and performance optimisations), and... I feel that's not enough, but I'm not sure what to add to that.
I'm running on very little budget (I just left school and I'd rather not go into my limited savings over this), so I can't afford to just throw money at ads.
Getting my historical fiction novel published. I finished writing it a couple of months back. Extremely short plot summary: guy deserts from the Roman XI legion, goes back home only to find that his entire province is about to revolt against the Roman Empire at the height of its power. It's one of the most spectacular feats of collective self-immolation in human history and it had a large effect on human history.
I think not enough has been written about the role of abject stupidity in human affairs. This book is an attempt to correct that.
Thanks! It's historical fiction so 90% true :-) The novel is set during the First Jewish–Roman War[0] and I tried as much as possible to adhere to what we know of actual events. There's a good case to be made that the effect it had on Judaism greatly impacted the development of Christianity. Plus it helped crown the Flavian Dynasty. I have just a short blurb and the first chapter here[1]. I'll definitely post more about the book and the process if I manage to get it published. There are quite a few aspiring novelists here so it might be encouraging to them.
Open source doorbell camera software running on Raspberry Pi Zero. Face recognition and motion detection already built-in, so I've set it up so that I get a message on element when someone is at the door and a video of any motion longer than 5 seconds. Not fully optimized yet.
I'm working on my hobby operating system written in Rust. It is completely text-based, but the console was lacking a scrollback buffer until this week. It's a simple feature, really, but having to redirect anything that outputs more than one screen to a file to read it was a pain. I'm happy to finally have it!
This weekend, I also made good progress on user-space memory and found a workaround for some issues I had. I still need to implement it the right way, though. After a few years on the project, the thing that is giving me the most trouble is grokking the concept of page tables.
I spent a week building a website that is a repository of timeless content. It generally consists of old magazine articles (though not exclusively) that you can pick up at any time and will bring value to your life. I notice most of what we consume on daily basis is ephemeral and not all that valuable outside of the moment, whereas ever-present content is always valuable.
The unique angle here is that each article includes a hand-written summary explaining why the article was meaningful to the curator. This gives you a quick window into the piece without being overwhelming.
Since this is a free website mostly to be shared with friends and family, I implemented user login via "phone number" to save and submit articles, but without a one-time token. So it's "password-less" for now; a trust-based system.
It's basically 'done' - but I'm not sure that I want to share it publicly for the aforementioned reasons. I've been using it on my subway rides to read more interesting stuff. So far so good.
From that I’m guessing you don’t want to open it to random HN folk, but if you ever do, it sounds a great resource and something I’d love to participate in.
Building algorithmic trading models. So far results continue to be good with every model outperforming the market on both absolute and risk-adjusted basis since going live.
This is an unleveraged, apples to apples comparison. These are not high frequency trading models. Most of them only change signal once every 2-4 weeks on average. During long signals, the models are simply long the S&P 500 and during short signals, they go to cash.
One of the pros of this macro swing-trading/hedging style is high tax efficiency, by holding a core ETF long position that never gets sold and then selling S&P 500 futures (ES or MES) of equal value to the ETFs against the long position. This way your account will accumulate unrealized capital gains indefinitely and you'll only pay tax on the net result of successful hedging. The cherry on top is that the S&P 500 futures are section 1256 contracts that are taxed at 60% long term / 40% short term capital gains rates regardless of the duration they are held.
The models use a variety of indicators, many of them custom built. Most important are various VIX metrics (absolute level, VIX futures curve shape/slope, divergences against S&P 500 price, etc), trend-following TA metrics (MACD, EMV, etc), mean-reversion TA metrics (Bollinger Bands, CMO, etc), macroeconomic (unemployment, housing starts, leading composite), and monetary policy (yield curve inversion, equity risk premium, dot plot, etc). They've been backtested very cautiously to avoid overfitting to the best of my ability.
I've been curious about doing algotrading for both the data engineering aspect and the quant. Do you have suggestions about books or others sources to get inspiration from ?
Is this a one man venture or do you have a group discussing edges ?
Building a little CLI tool that stores API requests and spins up a small REST server that allows them to be pinged. This stemmed from being on a flight, not wanting to buy WiFi, or the WiFi being slow and I just want to build something around an API.
Building an electric 1973 Ferrari Dino to compete in the battle of the builders at SEMA in November. Lots of fun electronics and software, as well as mechanical challenges to overcome.
I am working on https://geobase.app/ which is a platform for geospatial full-stack developers.
We have created workflows that a specific to the geospatial, mapping and GIS industry use cases. This is currently in private beta but going live in a few weeks. It is built on top of supabase's self-hosted stack.
I'm building up a small but resourceful artificial intelligence research group focusing on specializing in the triangle "machine learning - search - natural language processing".
Have got a bit of funding, a building, an 1.3 M€ GPU cluster. Also looking for Ph.D. candidates and contract developers.
(The hard part is spending the money wisely but in 8 weeks - it is a time-limited government budget that "expires" - while teaching writing papers and writing grant applications.)
Wow, just had to comment because this one is hilarious. Everyone else has the typical "I built my own accounting CLI" or "I'm thinking about maybe publishing my devtool", and you're over here casually describing your building. I just have to say: well done, that sounds like quite the life! The world will no doubt appreciate your toil one day, even if you feel stressed in the short term. IDK why, but your comment makes me want to share one of my favorite quotes:
Only with time will the period of my real influence begin and I trust that it will be a long one, for I am firmly convinced of Seneca's promise: "Although envy imposed silence on all who lived with you, those men will come who will judge without ill-will and without favour."
- Schopenhauer's Doctoral Dissertation, The Fourfold Root
I'd throw my hat in the ring as a philosophy-minded SWE who's coming up on the end of my runway while writing my book on unifying symbolic+connectionist AI (going for ~1y now), but your choice of currency tells me you probably don't have a need for any of us yanks. Instead, I'll send you my very strongest best wishes from across the ocean! And while I'm at it, I'll endorse my absolute favorite paper ever written on search, in case it sparks some ideas: Simon & Newell's Human Problem Solving (1970) https://github.com/vlall/Ai-Papers/blob/master/1971_Human%20... . If you're not already teaching it, ofc ;)
This quote in particular pops into my head at least once a day:
The problem solver's search for a solution is an odyssey through the problem space, from one knowledge state to another, until his current knowledge state includes the problem solution.
I am busy rebuilding my music studio with only gear from the late 90s/2000. It's much more fun to relive my youth and with patience less expensive than I thought.
Oh those were golden years in the studio, hope you have a lot of fun.
Just casting my mind back to the setup we had in '99:
Juno 106, Korg Monopoly, TX816. Prophet 5, Seq Circuits Pro1,
Novation bass station, 909, Korg Wavestation (rack), Korg M1, EMU
sampler (EIII rack IIRC), Atari Mega 4 + Cubase for seqs, a IoMega zip
drive for storage that killed every other project with click of death,
and fucking midi cables and audio snakes running everywhere... great
days.
Started my masters at Georgia Tech [1] this summer. Still in the first semester, but really enjoying so far. Just not much spare time on nights and weekends, but it helped me cut down the time spent on social/news/etc.
Thanks! Loving it so far. It has been a long time since I graduated, so using the first semester to get used to a regular study schedule again.
The program is massive (like, 1500+ students in my ML4T [1] class alone; the largest in GT's history), but academically rigorous, with a supportive community, active discussion forums, cool projects, and a small army of TAs. And very affordable, compared to all other programs out there.
It takes a long time for them to process applications, but they seem to honor their deadlines. I got my invitation a few days before the "you should hear from us until this date", and know other folks were on the same boat.
My primary project the past couple weeks has been applying the AI interpretability technique from Anthropic's famous paper this past summer (you know, the Golden Gate Claude one) to single-cell RNA-seq data. What works on one huge, inscrutible vector space ought to work on another, right? In either case, a fun way to keep in practice, both for comp bio and deep learning.
My side-project, in essence, the bash '&' operator for cases where the first process is already running. It took me months of searching before I could believe that this doesn't already exist, but there you go. I gave in to feature creep, of course, so it's a bit more than that now (I made a ncurses-based dashboard? Why??) but someday soon I'll make it public.
I may be misunderstanding what you're trying to achieve, but you can simulate the bash '&' operator for a running process by pressing ctrl-Z to suspend the process and send it to the background, then running 'bg' to continue the process in the background.
The few times I built TUIs with ncurses I wondered: why do I have to program so much by myself?
ncurses is so basic, I didn't have too much fun building UIs with it (more than once). Is there a wrapper or a more modern alternative out there that provides containers and widgets like most GUI frameworks do?
CV pipeline to get some additional realtime stats for an annual Mario Kart 8 LAN tournament my friends and I run, hoping to be able to get real time race position tracking and other stats like boost/drift %, per-player item distributions, average time to 10 coins, and whatever else we can think of (and make work)
You're going into first place too early. You kinda have to cruse in 2nd until the very end to avoid blue shells. Better yet, stay in 9th as long as possible - that's the highest position you can get the good drops like Blue Shells, Bullets, etc. But I think you can get red shells in 2nd, so just save those till the end.
Google Sheets add-on for quickly generating forecasts on seasonal, time series data: https://www.youtube.com/watch?v=nPUdTCxURgg . Submitted it to the Google Workspace Marketplace a few days ago, hoping it'll get approved soon!
I'm working on a foresight and emergency response team to see large calamities coming before they happen, and hopefully be able to do something about it. We put out weekly minutes at https://blog.sentinel-team.org
- Giving up Arma3 gamemode dev maintenance for a large-ish community, but hanging around to teach people git/github and provide wisdom / teach how to do software dev (a lot of folks have minimal software experience). Although debating whether to straight up leave the community.
Working on validating a startup idea I had 12 years ago. It's like Pingdom for ads, periodically checking if your ads are being blocked by AdBlockers.
I always thought the idea was somewhat weak, but not enough to discard entirely. So, along with a friend, I built a prototype over the last two weeks, and now we're trying to validate it: https://scanningfox.com/
I'm enjoying using Elixir for this project. As a long-time Erlang dev, I was initially skeptical about Elixir, but Phoenix.LiveView has changed my opinion.
Dance calendar (Django app, with some AlpineJS/HTMX). Approximately 80 admin pages + 20 end user views + 100 admin views. 2M page loads/month during the summertime. This is in production.
For the dance calendar, support site with good first-line AI support based on FAQ answers.
Ad management platform for the dance calendar. Sites like this have specific requirements for placing ads that are not well supported by AdSense etc. I would like to have an alternative for smaller players for the header bidding used by the larger players.
Separate webapp to store happenings (messages, emails, descriptions, documents, etc.), tag them and show them on a timeline, allowing filtering the events visible on the timeline. Django/HTMX/AlpineJS. This is for a legal battle I am having.
A tool for describing workflows using the Unified Service Management (USM) model. The method is to frameworks (ITIL etc.) what open source is to commercial software. I am currently working on cross-referencing tool to map ISO 27k requirements to USM statements. I have developed my own formal language for defining the requirements. The end goal is to automate validating many ISO 27k requirements.
Reverse engineering my e-bike’s head unit and motor controller to build a custom head unit out of a Raspberry Pi with oled touch screen (head unit will also be used to control LED patterns on the bike).
Used a logic analyzer to work out the protocol between the head unit and the motor controller (uart at 9600) and used a ESP32 to man in the middle the protocol. Currently reverse engineering the meaning of the bytes in the packets sent between the units.
First attempt was taking apart the head unit and attaching a debugger to the exposed serial debug interface (Cortex M0) chip, but looks like the manufacturer had disabled flash reading by setting the flash security bit.
Finally learning piano and drums after playing guitars for years. Thinking of own musical voice for first song, EP, and album. Made this chords poster to help: https://hypertexthero.com/piano/
Working on a default home.html page for my web browser with most used links, note pad, drawing pad, and forms for quickly creating posts for static sites that I’ll publish shortly. It would be nice if Firefox let you define a custom page for new tabs as well as new windows, instead of only Blank Page and Firefox Home (Default).
Usually have the relationship between work and play in mind, and how many of my favorite games have elements of compounding interest in rogue lite game modes where a little bit goes a long way with saved progression.
Thinking about the difficult, important work of nurses and caretakers while helping to manage care for an elderly relative.
I just started building a small website for structural engineers with various tools on it (eg. capacity of a steel column). I have spent thousands of hours in the past building fancy excel (incl. vba)and mathcad documents for personal use and this is my first time trying to do it with "real" code. I went with Blazor and c#, so far it looks like a good choice. The long term goal if the projects is a success and becomes popular would be to have a FEM engine for 2D frame structures running in the browser client-side.
It lets you use whatever you’re already doing for auth, and lets you become an oauth provider, or issue tokens instead of people passing API keys, on top of that.
The more interesting aspect of that you could use it to bolt on an entire app store or ecosystem on top of your existing product or api.
(ping me if you want to look at this more seriously from a business perspective!)
I made a website where you can play chess puzzles against Stockfish.
I had the idea for a website where you can play chess puzzles, but if you make the wrong move, the puzzle turns into a game against Stockfish. This opens the door to either find alternative solutions or fail miserably (at some point you realize you are not following the puzzle any more). I think its a more engaging way to play puzzles!
This is my first online project, feedback is highly appreciated!
I just picked up an old Sheldon Lathe. I got it for $200 and hauled it myself with a drop trailer. Its going to need a lot of work but its a beautiful old machine.
Currently shipping some updates to https://storywatch.org, think of us as Goodreads or IMDB for web fiction and fanfiction, rather than traditional dead tree books. If you are fans of books like Worm or HPMOR, give us a try.
If you find you're spending too much time on hacker news, you can actually set anti-procrastination settings in your profile. I use them and it's definitely helped a lot!
1. SPC kit [0]. Once made it to the front page! [1]
It's an SQL library for doing statistical process control (SPC) calculations.
This has been a labour of love for about 2 years now. I work on it sporadically. Recently I got more disciplined about what I am working on and I am slowly closing the gap on a first 0.1 release.
2. Finding work. As much fun as it is to tinker, I am nursing the standard crippling addiction to food and shelter. I am also nursing an increasing loathing for LinkedIn and wish to be free of having to check it.
I'm working on an application that I'm writing for myself (no release ambitions for the time being) that focuses on spaced repetition applied to chess.
It's similar to chess puzzles but with a twist: positions and moves are explained, there's a wider variety of exercises (such as excluding all the bad moves explaining why, improving the board position, and many others).
Just a fun little CRUD app built with Next.js, MUI, Prisma Postgres. I'm adding Halloween sketches now, if you know some good ones feel free to add them, or anything else :)
A custom lightweight insulated hard-sided truck camper for mid-size and half-tons. Decided against fiberglass since it can be a pain in the butt, especially with no garage and in the cold. Ideal would be no wood and minimal framing, but i'm not a mechanical engineer, so it's hard to calculate the forces involved, so over-building feels necessary. If anyone is a mechnical engineer, and bored, and would like to contribute their skills, I really want to open source the result so anyone can build it, using basic parts you can find at the big box store. So far I have a crappy model in FreeCAD and a lot of research material.
I've been messing around with a graph-based interface for task management over in https://github.com/crockeo/ekad. The interesting branches are:
- `main`, which currently houses a custom interactive graph visualizer built on top of the great `vello` from linebender (https://github.com/linebender/vello).
- `ch/typescript`, which has my attempts at joining a more traditional task manager with a graph visualization.
Instead of replacing the entire document or selection, we want it to create diffs or operations for the minimal amount of edits as possible. This helps preserve intent better when merging the doc later on with OT/CRDTs. (Of course, you could also ask GPT to semantically merge docs for you haha).
So far, it's been harder than plain text or spreadsheets which have an easier position/coordinate system to work with: just line-col or row-col.
I also have a few screen recordings showing its capabilities. I’ve been using it for about six months to enhance my writing.
The plugin leverages a language model to suggest text improvements, features a split-view interface, and allows users to select the edits they want to keep from a diff.
It’s still a bit rough around the edges, and the code is quite messy since I’m still learning Lua and the Neovim API. However, I’m gradually improving it whenever I find the time.
Working on my side project https://gametje.com. It's an alternative to Jackbox games but it has a lower barrier for entry. All you need is a device with a web browser to play (also works with Chromecast). I'm trying to write some better tutorials for the early games I created to make it easier to get a feel for the gameplay. Also trying to work on the ui/ux flow to make it easier for people to understand what it is and how to create and host a game.
It’s a pretty bad hack of HTML, CSS, and some JavaScript, saved by Laravel, trying to print slightly larger than A5 paper for a major on demand publisher standard paperback, with browser Print to PDF settings, and is unwieldy but to the point, has an ISBN as well as being NP-complete. Discovered for myself that the number of permutations and derangements of the same length are related by the ratio of Euler’s Number e.
Made my own company that provides services and consultations in drones, robotics, and even cybersecurity. Very slow business at this stage, if someone is in the same field or went through the same stage, any tips are welcome.
Easy Data Transform v2 ( https://www.easydatatransform.com ). It is GUI based data wrangling tool for people who aren't (or don't want to be) programmers. 20 years of running a 1-man software company in January.
I've been experimenting with customizing Postgres to run on top of FoundationDB, which to me would be the dream combination of Postgres' top-notch feature set and ecosystem with FoundationDB's unique resilience, scalability and transactional guarantees.
Haven't written up anything about it or published any code yet, but it's working pretty well and I haven't even had to fork Postgres with all the extensibility it offers!
My email is in my profile if anybody would like to chat about it
On writing some blog posts about things I’ve built lately (both at work and in my own time). Helps a lot with diving more deeply into topics than what’s reasonable for a “just needs to work” implementation.
I'm currently trying to implement an in-editor sandboxing/modding solution for the Godot game engine. It's hard work trying to make everything work the way people are used to having it, and even competing with GDScript.
Org-mode grammar (PEG parsing) for GNU Guile. Still at the very beginning, but hoping, that over time I can add more and more things, to make it a useful library for things like a minimal static blog based on org mode files or if any git hoster wants to have a good parser for org mode files to render readmes ...
That, and my personal website, using only HTML and CSS, and trying to keep it minimalistic, yet nice looking.
I'm working on an open-source event tracking infrastructure based on AWS (think Heap or Mixpanel but all infrastructure is in your AWS account and you own the data). It's incredible how much can be done just by combining AWS services.
A basic CRUD app around goal setting as a test for how AI tools can help write web apps to see how a modern web team could leverage this stuff to go faster, and maybe identify some missing pieces I could build one day. ChatGPT, Claude, Cursor, Copilot etc are genuinely great if you can already write code. They really let you blast straight through the mundane bits and focus on the hard stuff.
Working on an extension to my recently purchased outdoor AirGradient unit to add an atmospheric pressure sensor, a second temperature/humidity sensor, some connections to an external rain gauge and a way to power the unit without going through USB.
Many of the sensors and connections are small enough that I could have spun another PCB to replace the VOC/NOx module it comes with[1], but SGP41 ain't cheap & I wouldn't dare to desolder one from the existing module. So instead I'm going to try to use the extension I/O connector AG board has. Am currently waiting for my PCBs to arrive.
Speaking of PCBs. It is wonderful that it is possible to get 5 units of a prototype for a price of a coffee or two.
We’ve experimented with various approaches to promotion, including HN, KDP, Amazon Ads, and most recently Reddit Ads. It's been interesting to see which strategies resonate with the audience, but we're still figuring out the best way to get it in front of the right people.
Thanks for the questions! I hear you it does sound like a generic book name. Well, we have the domain and we couldn't really name it to something else as we think software engineering has many elements and we wanted to cover them.
Our book isn't just a technical book on software development. Instead, it goes into the life aspects of being a software engineer such as migration and parenting.
Many of us have wished for mentors who could guide us beyond the technicalities, offering insights into personal growth and career navigation. Recognizing this gap, we've created a resource that provides practical wisdom.
By taking a holistic approach to software engineering, we address both personal and professional development in a way that few other books do. This unique blend sets our book apart, offering a clear differentiator that defines our brand.
Any resources you'd recommend to learn marketing (esp. comming from a software background)? Asking for software friend who's having trouble marketing his software business.
Sounds really cool! Do you have a page describing your project with some pictures? It's funny I checked your post history before asking and found that you made the same comment I'm doing today to someone doing something similar one year ago! Is it what got you started?
Helping React devs move their state to the URL, in a type-safe way.
nuqs [1] started as a Next.js-only library, but recently I've been working on supporting all major React frameworks & routers (Remix, React Router, plain React with Vite etc).
iOS game to learn rhythm and drums [0]. It's MIDI based. Midi files to use as the tracks to practice on and Midi controllers to use as input. Here's a demo with my son on electronic drums [1]
Basically a mobile app with an integrated IDE and terminal with custom commands tailored to execute js code that interfaces with native android features
I’ve been building a copilot for an underserved language, and paused that in March with a little time since spent making a full language service: something where you can parse and resolve methods and types, and generally query for useful info. Perhaps it’s the root of a LSP server in future.
Recently got exicited about transforming my Tensorflow.js Parking mointoring application to a more general webapp that can do many things with Computer Vision, as well as upgrade from YOLO7.
Adding optional user sign-in to my word replacer browser extension: https://github.com/dan-lovelace/word-replacer-max. This is a precursor to adding generative replacement suggestions based on search terms. Even further, I'd like it to eventually analyze and de-trigger/disarm the copy on websites more broadly without having to define specific words or phrases.
A real-time interface for my Fantasy Premier League… league. It calculates point totals up to the minute, which the official app lacks for some reason (can take hours for final point tallies).
It’s mostly an excuse to play with data processing with duckdb, remote APIs, and Pocketbase.
A tool that makes it easier to run Docker Compose projects on NixOS. It’s essentially a Compose backend that targets a mix of NixOS + systemd + Podman/Docker.
Working on https://aliveai.app/. It is a simple wrapper frontend around a complex ComfyUI workflow. There is a lot of different things you can generate with StableDiffusion but I wanted to make it as easy as possible to create photo-realistic images of people.
The App is mostly being used for generating NSFW images though (which is ok).
I'm just dipping my toe into the Typescript water, with a rehype plugin that helps me turn markdown into <figure><img><figcaption></figcaption></figure> HTML.
The code's done, the yak shaving of packaging it as an npm module continues.
https://www.videototextai.com/ - an AI transcription, translation, chat with your video/audio platform. We are very close to releasing an update where it is possible to caption any video in any language - perfect for making social media content.
I'm developing a scuba diving app for the Apple Watch Ultra. It's an algorithm and graphical user interface that provides critical real-time information to recreational scuba divers while scuba diving. (Nautosys.com)
Gnosi: https://www.gnosi.ai/ . Y
You can auto-generate a constantly updated encyclopedia from a set of documents and conversations with an AI about those documents. Trying to make the process of having a Zettlkasten or personal notetaking system frictionless.
Working on https://www.brief.news - a completely personalized daily newsletter on the topics you're interested in. We've just launched the ability to add custom topics, so you can create a newsletter on anything now!
torchtune (https://github.com/pytorch/torchtune) - a PyTorch library for fine-tuning LLMs, particularly for memory-constrained setups. Try it out and fine-tune Llama3.1 8B on a single RTX 4090!
fond memories. we did tv guides for a local cable system for a bit in the early 90s. they had a customer db, but no "export" beyond sending print jobs to their bigarse line printer.
but it was a parallel port and i had use of a luggable with a bidirectional parallel port so i'd haul that in once a month, hook it up, and have them run a "hello customer" fake billing run, which was hoovered up and stripped down to the address list we needed to mail to that month.
Working on my first foray into DML-based speakers. Low expectations, but more enjoying the fun of learning a new domain (and it’s a distraction from building garden furniture). Got various exciters, panel types, mount ideas ready, just taking the time out of the evenings.
Anyone know of a community where you genuinely try each other’s stuff and give feedback? Sometimes lose motivation aimlessly guessing what people want without having users to give actual feedback.
I’m working on a replacement for file coreutils (touch, cp, mv, rm) which prompt before deleting by default and have a more sane api overall. Intended for interactive use on the CLI. In Rust.
Building a narrative interactive novel/RPG thing. Wrapping music composition this week, writing about half done, game engine built, art finishing around Christmas.
Working on a data exploration and visualization tool for SQLite. Some of the features include: ER diagram, browse table data, charting (bar, column, line, pie, scatter).
I'm currently working on a new blockchain technology called Roughchain. Earlier today, I shared the whitepaper in this HN thread [1], and I’ve already received some valuable feedback.
The core architecture is split into two components: a timestamping signing service and a P2P gossip network. By decoupling the gossip network, I'm simulating its performance using a Monte Carlo approach. With a basic gossip protocol, the simulation reaches ~10k TPS on a 100-node, randomly connected network (not fully connected), and I see a lot of potential for further protocol optimizations.
Initially, I considered a more resource-intensive approach using Shadow [2] for more realistic node simulations, as outlined in this discussion on libp2p's Gossipsub stress metrics [3]. However, the Monte Carlo method allows me to simulate the network more efficiently without needing to deploy full nodes.
In parallel, I'm exploring game-theoretical concepts for selecting signers and ensuring the system remains open to new entrants. One paper I'm currently diving into is "Collusion, Efficiency, and Dominant Strategies" [4].
I'm working on this food logging app / personal health dashboard that I'm building for myself.
I was asked to log my food by a dietician I'm working with, because I was constantly feeling hunger after cycling training.
But all the solution out there were quite complex to input data, and I just want to write down plain text with some additional markup and be able to generate some graphs and recognize some patterns out it.
Decided to do it T3 and throw in all those weird technologies just to see what's out there.
Diverse opinions and even conflict in small groups can be productive provided there is trust. I’d love to understand trust better. Got anything to share or places to start?
I’ve been on a deep dive on the philosophy of harmony for a long time. Just submitted my second major article on “harmony of opposites” that deals with the role of conflict/tension in harmony.
Alignment of interest and principles of conflict resolution and
diplomacy are where I got to with "dynamic trust systems" at the
moment. The whole project is to kinda push "beyond authentication".
Yes I'd love to share some as I've been seeking proof-readers in some
security communities. If you DM me via cybershow,uk (email in footer)
we can chat.
Both of these! And spot on with (mutual) motive analysis to come up
with an "alignment" matrix.
I was inspired some time ago by Stella Rimington's writing (ex MI5
chief) that "identity" is actually a very poor basis for trust and
authentication. It's unnerving watching the whole zero-trust show
organise itself around notions of strong identity (as opposed to role
and earned trust), which might turn out to be a rather silly thing to
do.
I'm still pretty early in terms of figuring out if this approach is better for larger projects. I can confidently say it works great on small things - for small changes on small files the AST approach works pretty well. You can say things like "add a click listener to the button that calls a function to tally the user's score" to a simple game and it will do it. That is, less than a minute after typing the prompt, you will see the code update in the editor and see your preview-webview update with your changes applied.
However, I have noticed that the AST code quality heavily depends on how common it is in the training set. I think I will have to add documentation to it through RAG or something - because OpenAI's models that I'm using seem to have limited experience writing esprima for JavaScript for example.
So it's hit or miss. In some cases I do feel like I'm throwing stupid compute at solving small problems and it's unnecessary - however, as I work on the project, it is getting better and better at successfully making the modifications. Some of that is me improving the prompts, some of it is OpenAI improving the models themselves, and some of it is the infrastructure I'm building for the project itself.
I did notice a huge improvement when o1-mini released. It is dramatically better at writing the AST code than GPT-4o or 4o-mini. I haven't tried Claude 3.5 yet but I've been hearing it does an exceptional job at code writing - not sure about my AST requirements though!
Lots of people have visited but a launch on HN isn’t enough on its own. I’m trying to figure out how to get the word out to more people to kickstart it. The goal is for it to be a community that people return to as part of their daily online life. That’s not a programming problem so it’s hard (for me).
I am working on (yet another?) DuckDB gui, with an emphasis on devops-y/back office workflows.
I want to join data from AWS with other sources and present a nice data-table UI, and perhaps allow taking some basic actions on a row, defining some filters, etc.
Have you ever tried to copy-paste data out of the AWS console? Truly a terrible experience.
On Mac, pbcopy and pbpaste don't work with images. I've written one that does in python, so I can right click on an image in Chrome, hit copy, then do pbipaste > foo.png and have it work. And then also do pbicopy < foo.png, and then be able to paste in Preview or whatever.
It seems that adding one node to a linked list causes a repeatable performance drop. That is, linked lists of say 2^21 nodes sort faster than lists of 2^21 + 1 nodes.
Trying to bring the finishing touches to a Common Lisp SSG I made during a handful of vacation afternoons (https://git.sr.ht/~q3cpma/make-website) and filling the resulting website (https://world-playground-deceit.net/) with more content. More motivated by #2 now that the generator does 99% of what I need.
I learned how to weld (MIG) and built a giant mushroom to house a mannequin I dubbed "the mushroom man" over about 100 hours in the last 4 weeks. I covered the outside with thick foam panels cut to size, cementing them in place with copious amounts of spray foam. I shaved the outside to a nice shape with a sawzall and the inside I covered in chicken-fenced, then attached a painters tarp to that (so it could be painted on).
To fit on a trailer (the mushroom's cap is 11.5ft wide) the cap comes off the stem and the edges of the cap are two half-moons which have fixed mounting points where threaded rod sticks through some welded washers, and a nut is put on in place. I was too last minute to install the 200 WS2811 pixels and have them run some cool patterns, before the music festival I brought it to came time, but even just a lantern on top (another painters tarp covered the cap's metal-frame, and everything was spray painted) looked great.
Super fun project. Expensive, but I learned a lot, got to be creative, and I'm happy to try out new things and make the best of my before-children time. Also, it was such a joy seeing people croud around the mushroom (and site beside the mushroom man inside) at night during the festival.
That's incredible - do you have any photos of this anywhere?
I learned to weld awhile back, but haven't pulled the trigger on purchasing all the stuff I need.
Tuilet: a TUI for Toilet, the premier ANSI text generator, written in Rust. https://github.com/gamache/tuilet
Are you an IRC shitposter? Isn't it hard to experiment with Toilet/Figlet fonts and flags? Well _not anymore._ Presenting Tuilet: a front-end to Toilet written by us, for us.
Built a site that will follow development of popular upstream open source projects on a daily basis and uses AI to summarize the commits and attempt to make it easier to track what's going on with the project: https://gitpulse.org
I’m working on https://www.zonehero.io/ to deliver a cost effective alternative to AWS native load balancing (ALB).
We already had a PoC which saved our customer 50% vs ALB (in their case that’s more than a million dollars a year), we’re now working on tooling and scaling the solution up!
Next we want to bring wasm modules to the lbs, to do edge traffic curation, bring your own model, etc.
I've been working on https://wut.dev in my spare time.
It's essentially a simpler, read-only, AWS dashboard where everything is a filterable, searchable, exportable-to-CSV table, with some extra features like multi-region mode, saved notes, and a debugger for access denied errors.
It uses the AWS SDK for JavaScript, so everything is run client-side from your browser. I'm not 100% sure what direction I'm taking it yet, but it's been fun to hack on!
There's a live demo here: https://wut.dev/?service=ec2&type=instances&demo=true if you want to try it out.
are you designing the UI for every service by hand, or doing it programmatically somehow?
Thankfully programmatic. It’s a common UI table widget, essentially, and I’ve written some custom code to handle multi-region support, updating the AWS credential handler, pagination, and response processing. From there, it’s a matter of plugging in some common options for each AWS service: the service name, SDK method to call, pagination property (annoyingly, AWS API has numerous ways of paginating responses), etc. Takes about five minutes to add a new service.
I'm working on a 2D computational geometry library with a focus on irregular part packing in Rust. IE laying out laser cuts for wood or water jet for metal. https://crates.io/crates/babushka
I've got some nice types set up and a no fit polygon algorithm. Working on the genetic algorithm for packing à là svgnest.
I‘m building https://sql-workbench.com in my spare time.
It‘s a SQL Workbench in the browser, based on DuckDB WASM. You can query remote and local datasources, such as CSV, JSON or Parquet files.
You can also visualize the results, and share the queries via URL. Let me know what you think!
Good timing, as I just put the finishing touches on this month's devlog:
https://www.patreon.com/posts/september-2024-113011369?utm_m...
Basically, my rhythm-based roguelike on original NES now has a proper economy, with gold gain and shops to spend the gold in. It also now supports PAL and Dendy systems, which is especially wonky due to the different framerate, but helped a bit by this being a rhythm game. As long as the music plays at the correct tempo, the rest of the game adapts its speed and "feels" correct at the lower framerate.
Tons of work left to do, most of it pixel art (I'm learning as I go) but it's progressing quite nicely.
I'm building a greenhouse. The frame is done, and I've got plastic and a door on it. Next, I'd like to build boxes to hold soil and allow for easy watering.
This sounds so nice!
Got any progress photos?
I'm working on https://nuenki.app/, a language learning tool. It teaches you a language while you procrastinate by inserting translations of appropriate-difficulty sentences into webpages as you browse HN etc.
Currently trying to reduce costs by switching from using DeepL (high quality, low latency, high cost) everywhere to a hybrid that also uses Claude (high quality, high latency, low cost) for text that is far from the user. Also experimenting with Gemma 2 9B via Groq to go in between them, but it's bad at following instructions and I don't quite trust the quality numbers I'm seeing for it (they're benchmarked with gpt-4o as a judge).
I'm also trying to work out marketing. I'm not good at it, and I dislike it, but I need to get good at it. Currently considering Reddit ads for awareness, some content marketing going over the technical details (there's some fun language processing and performance optimisations), and... I feel that's not enough, but I'm not sure what to add to that.
I'm running on very little budget (I just left school and I'd rather not go into my limited savings over this), so I can't afford to just throw money at ads.
This is awesome!
Getting my historical fiction novel published. I finished writing it a couple of months back. Extremely short plot summary: guy deserts from the Roman XI legion, goes back home only to find that his entire province is about to revolt against the Roman Empire at the height of its power. It's one of the most spectacular feats of collective self-immolation in human history and it had a large effect on human history. I think not enough has been written about the role of abject stupidity in human affairs. This book is an attempt to correct that.
That sounds really worth reading! Please post it on HN when it’s published (or do you have any links / info now?)
I know it’s fiction but the way you phrase it makes it sound like this might be inspired by a true revolution, is that the case?
Thanks! It's historical fiction so 90% true :-) The novel is set during the First Jewish–Roman War[0] and I tried as much as possible to adhere to what we know of actual events. There's a good case to be made that the effect it had on Judaism greatly impacted the development of Christianity. Plus it helped crown the Flavian Dynasty. I have just a short blurb and the first chapter here[1]. I'll definitely post more about the book and the process if I manage to get it published. There are quite a few aspiring novelists here so it might be encouraging to them.
[0] https://en.wikipedia.org/wiki/First_Jewish%E2%80%93Roman_War
[1] https://jonyomtov.me/the-deserter
Open source doorbell camera software running on Raspberry Pi Zero. Face recognition and motion detection already built-in, so I've set it up so that I get a message on element when someone is at the door and a video of any motion longer than 5 seconds. Not fully optimized yet.
I'm working on my hobby operating system written in Rust. It is completely text-based, but the console was lacking a scrollback buffer until this week. It's a simple feature, really, but having to redirect anything that outputs more than one screen to a file to read it was a pain. I'm happy to finally have it!
This weekend, I also made good progress on user-space memory and found a workaround for some issues I had. I still need to implement it the right way, though. After a few years on the project, the thing that is giving me the most trouble is grokking the concept of page tables.
https://moros.cc
https://github.com/vinc/moros
I spent a week building a website that is a repository of timeless content. It generally consists of old magazine articles (though not exclusively) that you can pick up at any time and will bring value to your life. I notice most of what we consume on daily basis is ephemeral and not all that valuable outside of the moment, whereas ever-present content is always valuable.
The unique angle here is that each article includes a hand-written summary explaining why the article was meaningful to the curator. This gives you a quick window into the piece without being overwhelming.
Since this is a free website mostly to be shared with friends and family, I implemented user login via "phone number" to save and submit articles, but without a one-time token. So it's "password-less" for now; a trust-based system.
It's basically 'done' - but I'm not sure that I want to share it publicly for the aforementioned reasons. I've been using it on my subway rides to read more interesting stuff. So far so good.
screenshot - https://i.imgur.com/8kIrgBt.png
From that I’m guessing you don’t want to open it to random HN folk, but if you ever do, it sounds a great resource and something I’d love to participate in.
For anyone that wants access just shoot me an email (it's on my HN profile).
Building algorithmic trading models. So far results continue to be good with every model outperforming the market on both absolute and risk-adjusted basis since going live.
Since launching https://grizzlybulls.com in January 2022:
Model | Return | Max drawdown
-------------------
S&P 500 (benchmark) | 21.51% | -27.56%
VIX TA Macro MP Extreme | 64.21% | -16.48%
VIX TA Macro Advanced| 59.13% | -19.12%
VIX TA Advanced | 35.20% | -22.96%
VIX Advanced | 33.39% | -23.93%
VIX Basic | 24.29% | -24.23%
TA - Mean Reversion | 22.30% | -19.92%
TA - Trend | 27.07% | -24.98%
This is an unleveraged, apples to apples comparison. These are not high frequency trading models. Most of them only change signal once every 2-4 weeks on average. During long signals, the models are simply long the S&P 500 and during short signals, they go to cash.
One of the pros of this macro swing-trading/hedging style is high tax efficiency, by holding a core ETF long position that never gets sold and then selling S&P 500 futures (ES or MES) of equal value to the ETFs against the long position. This way your account will accumulate unrealized capital gains indefinitely and you'll only pay tax on the net result of successful hedging. The cherry on top is that the S&P 500 futures are section 1256 contracts that are taxed at 60% long term / 40% short term capital gains rates regardless of the duration they are held.
The models use a variety of indicators, many of them custom built. Most important are various VIX metrics (absolute level, VIX futures curve shape/slope, divergences against S&P 500 price, etc), trend-following TA metrics (MACD, EMV, etc), mean-reversion TA metrics (Bollinger Bands, CMO, etc), macroeconomic (unemployment, housing starts, leading composite), and monetary policy (yield curve inversion, equity risk premium, dot plot, etc). They've been backtested very cautiously to avoid overfitting to the best of my ability.
I've been curious about doing algotrading for both the data engineering aspect and the quant. Do you have suggestions about books or others sources to get inspiration from ?
Is this a one man venture or do you have a group discussing edges ?
If so good why are you selling it as a service ?
After a long break, I'm working on writing about Japan things I find interesting: https://onefromnippon.com/
I'm hoping Japanese vending machines will be an interesting topic for the next post.
Building a little CLI tool that stores API requests and spins up a small REST server that allows them to be pinged. This stemmed from being on a flight, not wanting to buy WiFi, or the WiFi being slow and I just want to build something around an API.
I called it gofaux: https://github.com/tjb/gofaux
That sounds potentially really neat for mocking API calls.
Hey! I'm working on Getgud.io, an AI-powered game analytics and anti-cheat platform.
Our goal is to provide complete observability into player behavior, detect cheaters and griefers, and help game developers improve player retention.
Some key features we're working on:
- AI-powered analysis of in-match player actions to detect anomalies
- Customizable rules engine for automated responses to toxic behavior
- Visual replay system for reviewing flagged matches
Check out our website at https://www.getgud.io and watch our detection video at https://www.youtube.com/watch?v=5EhTpfEzh1M to see Getgud.io in action.
We support server-side integration for popular platforms like Unreal Engine and Unity.
For integration guides and SDK references, visit our docs at https://github.com/getgud-io/getgud-docs.
Happy to chat more about game analytics and cheat detection if anyone's interested!
Building an electric 1973 Ferrari Dino to compete in the battle of the builders at SEMA in November. Lots of fun electronics and software, as well as mechanical challenges to overcome.
I am working on https://geobase.app/ which is a platform for geospatial full-stack developers.
We have created workflows that a specific to the geospatial, mapping and GIS industry use cases. This is currently in private beta but going live in a few weeks. It is built on top of supabase's self-hosted stack.
We were recently also featured on motherduck's blog https://motherduck.com/blog/pushing-geo-boundaries-with-moth...
I'm building up a small but resourceful artificial intelligence research group focusing on specializing in the triangle "machine learning - search - natural language processing".
Have got a bit of funding, a building, an 1.3 M€ GPU cluster. Also looking for Ph.D. candidates and contract developers. (The hard part is spending the money wisely but in 8 weeks - it is a time-limited government budget that "expires" - while teaching writing papers and writing grant applications.)
Wow, just had to comment because this one is hilarious. Everyone else has the typical "I built my own accounting CLI" or "I'm thinking about maybe publishing my devtool", and you're over here casually describing your building. I just have to say: well done, that sounds like quite the life! The world will no doubt appreciate your toil one day, even if you feel stressed in the short term. IDK why, but your comment makes me want to share one of my favorite quotes:
- Schopenhauer's Doctoral Dissertation, The Fourfold RootI'd throw my hat in the ring as a philosophy-minded SWE who's coming up on the end of my runway while writing my book on unifying symbolic+connectionist AI (going for ~1y now), but your choice of currency tells me you probably don't have a need for any of us yanks. Instead, I'll send you my very strongest best wishes from across the ocean! And while I'm at it, I'll endorse my absolute favorite paper ever written on search, in case it sparks some ideas: Simon & Newell's Human Problem Solving (1970) https://github.com/vlall/Ai-Papers/blob/master/1971_Human%20... . If you're not already teaching it, ofc ;)
This quote in particular pops into my head at least once a day:
Hey guys! We're engineers/designers from France, and we've built the Ultimate DIY Battery that you can repair and refill!
- Ride Sustainably with the World's First Repairable Battery
- Refillable in 5 minutes (just buy $150 worth of new cells every 3 years or so, when they're depleted)
- Be Worry-Free thanks to the Fireproof Casing! There's been waaaaaay too many lithium fires!
It's launching as an IndieGogo in one week and there is an offer for early-backers here https://get.gouach.com/1 for a 25% discount on the battery!
btw do you plan to leverage your pcb connector plates for other kinds of batteries ?
I am busy rebuilding my music studio with only gear from the late 90s/2000. It's much more fun to relive my youth and with patience less expensive than I thought.
Oh those were golden years in the studio, hope you have a lot of fun. Just casting my mind back to the setup we had in '99:
Juno 106, Korg Monopoly, TX816. Prophet 5, Seq Circuits Pro1, Novation bass station, 909, Korg Wavestation (rack), Korg M1, EMU sampler (EIII rack IIRC), Atari Mega 4 + Cubase for seqs, a IoMega zip drive for storage that killed every other project with click of death, and fucking midi cables and audio snakes running everywhere... great days.
ADAT?
Started my masters at Georgia Tech [1] this summer. Still in the first semester, but really enjoying so far. Just not much spare time on nights and weekends, but it helped me cut down the time spent on social/news/etc.
[1] https://omscs.gatech.edu/
Good luck! How is it going? How long did it take for them to accept you? Asking because a friend applied in July and still not heard back from them.
Thanks! Loving it so far. It has been a long time since I graduated, so using the first semester to get used to a regular study schedule again.
The program is massive (like, 1500+ students in my ML4T [1] class alone; the largest in GT's history), but academically rigorous, with a supportive community, active discussion forums, cool projects, and a small army of TAs. And very affordable, compared to all other programs out there.
It takes a long time for them to process applications, but they seem to honor their deadlines. I got my invitation a few days before the "you should hear from us until this date", and know other folks were on the same boat.
Hope your friend gets their invitation soon.
[1] https://omscs.gatech.edu/cs-7646-machine-learning-trading
My primary project the past couple weeks has been applying the AI interpretability technique from Anthropic's famous paper this past summer (you know, the Golden Gate Claude one) to single-cell RNA-seq data. What works on one huge, inscrutible vector space ought to work on another, right? In either case, a fun way to keep in practice, both for comp bio and deep learning.
My side-project, in essence, the bash '&' operator for cases where the first process is already running. It took me months of searching before I could believe that this doesn't already exist, but there you go. I gave in to feature creep, of course, so it's a bit more than that now (I made a ncurses-based dashboard? Why??) but someday soon I'll make it public.
I may be misunderstanding what you're trying to achieve, but you can simulate the bash '&' operator for a running process by pressing ctrl-Z to suspend the process and send it to the background, then running 'bg' to continue the process in the background.
The few times I built TUIs with ncurses I wondered: why do I have to program so much by myself? ncurses is so basic, I didn't have too much fun building UIs with it (more than once). Is there a wrapper or a more modern alternative out there that provides containers and widgets like most GUI frameworks do?
CV pipeline to get some additional realtime stats for an annual Mario Kart 8 LAN tournament my friends and I run, hoping to be able to get real time race position tracking and other stats like boost/drift %, per-player item distributions, average time to 10 coins, and whatever else we can think of (and make work)
Neat. I’d love something that could watch my wife and I play and confirm I really do get hit by more blue shells than her
You're going into first place too early. You kinda have to cruse in 2nd until the very end to avoid blue shells. Better yet, stay in 9th as long as possible - that's the highest position you can get the good drops like Blue Shells, Bullets, etc. But I think you can get red shells in 2nd, so just save those till the end.
Google Sheets add-on for quickly generating forecasts on seasonal, time series data: https://www.youtube.com/watch?v=nPUdTCxURgg . Submitted it to the Google Workspace Marketplace a few days ago, hoping it'll get approved soon!
I'm working on a foresight and emergency response team to see large calamities coming before they happen, and hopefully be able to do something about it. We put out weekly minutes at https://blog.sentinel-team.org
- Taking Octachainer and making my own CLI implementation to learn Rust https://www.elektronauts.com/t/octachainer-v1-3/45320
- hacking around with FunDSP (turns out it’s pretty fun), again to learn some Rust https://github.com/SamiPerttu/fundsp
- Giving up Arma3 gamemode dev maintenance for a large-ish community, but hanging around to teach people git/github and provide wisdom / teach how to do software dev (a lot of folks have minimal software experience). Although debating whether to straight up leave the community.
I'm working on https://www.ironcalc.com as a side project.
A spreadsheet engine with an open source permissive license. I have high hopes for it but I'm still in early stages of the project.
Ohh this is awesome! Thanks for sharing! How much time per week do you work on something like this?
I work a little bit on it every day. I guess somewhere around 15 hours per week. Glad you like it.
Working on validating a startup idea I had 12 years ago. It's like Pingdom for ads, periodically checking if your ads are being blocked by AdBlockers.
I always thought the idea was somewhat weak, but not enough to discard entirely. So, along with a friend, I built a prototype over the last two weeks, and now we're trying to validate it: https://scanningfox.com/
I'm enjoying using Elixir for this project. As a long-time Erlang dev, I was initially skeptical about Elixir, but Phoenix.LiveView has changed my opinion.
Liveview is so much fun. I want to build more things with it.
Dance calendar (Django app, with some AlpineJS/HTMX). Approximately 80 admin pages + 20 end user views + 100 admin views. 2M page loads/month during the summertime. This is in production.
For the dance calendar, support site with good first-line AI support based on FAQ answers.
Ad management platform for the dance calendar. Sites like this have specific requirements for placing ads that are not well supported by AdSense etc. I would like to have an alternative for smaller players for the header bidding used by the larger players.
Separate webapp to store happenings (messages, emails, descriptions, documents, etc.), tag them and show them on a timeline, allowing filtering the events visible on the timeline. Django/HTMX/AlpineJS. This is for a legal battle I am having.
A tool for describing workflows using the Unified Service Management (USM) model. The method is to frameworks (ITIL etc.) what open source is to commercial software. I am currently working on cross-referencing tool to map ISO 27k requirements to USM statements. I have developed my own formal language for defining the requirements. The end goal is to automate validating many ISO 27k requirements.
Reverse engineering my e-bike’s head unit and motor controller to build a custom head unit out of a Raspberry Pi with oled touch screen (head unit will also be used to control LED patterns on the bike).
Used a logic analyzer to work out the protocol between the head unit and the motor controller (uart at 9600) and used a ESP32 to man in the middle the protocol. Currently reverse engineering the meaning of the bytes in the packets sent between the units.
First attempt was taking apart the head unit and attaching a debugger to the exposed serial debug interface (Cortex M0) chip, but looks like the manufacturer had disabled flash reading by setting the flash security bit.
Finally learning piano and drums after playing guitars for years. Thinking of own musical voice for first song, EP, and album. Made this chords poster to help: https://hypertexthero.com/piano/
Working on a default home.html page for my web browser with most used links, note pad, drawing pad, and forms for quickly creating posts for static sites that I’ll publish shortly. It would be nice if Firefox let you define a custom page for new tabs as well as new windows, instead of only Blank Page and Firefox Home (Default).
Usually have the relationship between work and play in mind, and how many of my favorite games have elements of compounding interest in rogue lite game modes where a little bit goes a long way with saved progression.
Thinking about the difficult, important work of nurses and caretakers while helping to manage care for an elderly relative.
I just started building a small website for structural engineers with various tools on it (eg. capacity of a steel column). I have spent thousands of hours in the past building fancy excel (incl. vba)and mathcad documents for personal use and this is my first time trying to do it with "real" code. I went with Blazor and c#, so far it looks like a good choice. The long term goal if the projects is a success and becomes popular would be to have a FEM engine for 2D frame structures running in the browser client-side.
I built a service that lets you bolt an oauth/oidc provider onto your app with a single callback. https://github.com/poundifdef/connectivly
It lets you use whatever you’re already doing for auth, and lets you become an oauth provider, or issue tokens instead of people passing API keys, on top of that.
The more interesting aspect of that you could use it to bolt on an entire app store or ecosystem on top of your existing product or api.
(ping me if you want to look at this more seriously from a business perspective!)
https://www.chesspuzzlebot.com/
I made a website where you can play chess puzzles against Stockfish.
I had the idea for a website where you can play chess puzzles, but if you make the wrong move, the puzzle turns into a game against Stockfish. This opens the door to either find alternative solutions or fail miserably (at some point you realize you are not following the puzzle any more). I think its a more engaging way to play puzzles!
This is my first online project, feedback is highly appreciated!
I just picked up an old Sheldon Lathe. I got it for $200 and hauled it myself with a drop trailer. Its going to need a lot of work but its a beautiful old machine.
Writing up a paper for my PhD on machine learning with non-Euclidean loss functions.
Taking over the affairs of one of my elderly relatives now that she can't manage by herself.
A project with a medical insurer for adjudicating insurance claims using LLMs.
Currently shipping some updates to https://storywatch.org, think of us as Goodreads or IMDB for web fiction and fanfiction, rather than traditional dead tree books. If you are fans of books like Worm or HPMOR, give us a try.
Working on actively trying to quit reading HackerNews.
If you find you're spending too much time on hacker news, you can actually set anti-procrastination settings in your profile. I use them and it's definitely helped a lot!
One way that worked for me in trying to quit Reddit was to edit /etc/hosts, and redirect it to localhost.
Limiting my Reddit intake this way actually worked, but I don't feel the need for HN yet.Start by creating an AI to read it for you and summarize the discussions, so you have less to read.
Is it open-source?
Try silent quitting HN, or living with your eyes closed! Worked for me, for 5 mins.
I see you are making strides
I’ve found if I announce I want to make a change I’m more likely to follow through. Maybe they are trying it out?
Why?
1. SPC kit [0]. Once made it to the front page! [1]
It's an SQL library for doing statistical process control (SPC) calculations.
This has been a labour of love for about 2 years now. I work on it sporadically. Recently I got more disciplined about what I am working on and I am slowly closing the gap on a first 0.1 release.
2. Finding work. As much fun as it is to tinker, I am nursing the standard crippling addiction to food and shelter. I am also nursing an increasing loathing for LinkedIn and wish to be free of having to check it.
[0] https://github.com/jchester/spc-kit
[1] https://news.ycombinator.com/item?id=39612775
I'm working on an application that I'm writing for myself (no release ambitions for the time being) that focuses on spaced repetition applied to chess.
It's similar to chess puzzles but with a twist: positions and moves are explained, there's a wider variety of exercises (such as excluding all the bad moves explaining why, improving the board position, and many others).
Working on Perseid (https://perseid.dev), a framework that allows web developers to ship full-stack apps in minutes, using their favorite stack.
I'm working on a Sketch Comedy Database website:
https://www.sketchtv.lol/
https://github.com/swax/SCDB
Just a fun little CRUD app built with Next.js, MUI, Prisma Postgres. I'm adding Halloween sketches now, if you know some good ones feel free to add them, or anything else :)
A custom lightweight insulated hard-sided truck camper for mid-size and half-tons. Decided against fiberglass since it can be a pain in the butt, especially with no garage and in the cold. Ideal would be no wood and minimal framing, but i'm not a mechanical engineer, so it's hard to calculate the forces involved, so over-building feels necessary. If anyone is a mechnical engineer, and bored, and would like to contribute their skills, I really want to open source the result so anyone can build it, using basic parts you can find at the big box store. So far I have a crappy model in FreeCAD and a lot of research material.
I've been messing around with a graph-based interface for task management over in https://github.com/crockeo/ekad. The interesting branches are:
- `main`, which currently houses a custom interactive graph visualizer built on top of the great `vello` from linebender (https://github.com/linebender/vello).
- `ch/typescript`, which has my attempts at joining a more traditional task manager with a graph visualization.
Prompting GPT to do rich text editing.
Instead of replacing the entire document or selection, we want it to create diffs or operations for the minimal amount of edits as possible. This helps preserve intent better when merging the doc later on with OT/CRDTs. (Of course, you could also ask GPT to semantically merge docs for you haha).
So far, it's been harder than plain text or spreadsheets which have an easier position/coordinate system to work with: just line-col or row-col.
Rich text is usually stored as trees with json or html. Have seen a paper (https://www.inkandswitch.com/peritext/) that represents it as a flat array.
Difference in approach would then be: is it easier for gpt to work with diffs or with operations/tool calls?
I’ve been developing a plugin for the Neovim text editor called sia.nvim, inspired by the Egyptian deity Sia.
You can check out the GitHub repository here: https://github.com/isaksamsten/sia.nvim.
I also have a few screen recordings showing its capabilities. I’ve been using it for about six months to enhance my writing.
The plugin leverages a language model to suggest text improvements, features a split-view interface, and allows users to select the edits they want to keep from a diff.
It’s still a bit rough around the edges, and the code is quite messy since I’m still learning Lua and the Neovim API. However, I’m gradually improving it whenever I find the time.
wow could you please share more of this work? I've been wondering how to make a diff tool / simple text editor for academic writing
Stack Frontend: NextJS, Tailwind, NextUI, Rust (via wasm-bindgen) hosted on Vercel
Backend: Rust, Axum server, LMDB hosted on Alwyzon, Cloudflare for CDN caching and SSL.
https://www.arible.co A growing directory of useful productivity tools accessible without multiple subscriptions or registrations
Working on my side project https://gametje.com. It's an alternative to Jackbox games but it has a lower barrier for entry. All you need is a device with a web browser to play (also works with Chromecast). I'm trying to write some better tutorials for the early games I created to make it easier to get a feel for the gameplay. Also trying to work on the ui/ux flow to make it easier for people to understand what it is and how to create and host a game.
Working on a puzzle book series (for lovers, maybe) https://www.kakurokokoro.com
It’s a pretty bad hack of HTML, CSS, and some JavaScript, saved by Laravel, trying to print slightly larger than A5 paper for a major on demand publisher standard paperback, with browser Print to PDF settings, and is unwieldy but to the point, has an ISBN as well as being NP-complete. Discovered for myself that the number of permutations and derangements of the same length are related by the ratio of Euler’s Number e.
I am working on https://www.merklemap.com/ A subdomain / CT search engine.
I’m chipping away at bugs and adding features to my app Reminder Sync for Obsidian! https://turquoisehexagon.co.uk/remindersync/
Currently working on supporting dataview tasks format and multiple reminder lists.
The app supports any markdown backed notes app, but I fear I may have limited its appeal by including Obsidian in the app name.
Edit: previous discussion https://news.ycombinator.com/item?id=39764919
Made my own company that provides services and consultations in drones, robotics, and even cybersecurity. Very slow business at this stage, if someone is in the same field or went through the same stage, any tips are welcome.
do gov work and print money
https://github.com/polymonster/diig
A music digging app for record collectors, with instagram style feed for listening to new vinyl snippets
Easy Data Transform v2 ( https://www.easydatatransform.com ). It is GUI based data wrangling tool for people who aren't (or don't want to be) programmers. 20 years of running a 1-man software company in January.
I've been experimenting with customizing Postgres to run on top of FoundationDB, which to me would be the dream combination of Postgres' top-notch feature set and ecosystem with FoundationDB's unique resilience, scalability and transactional guarantees.
Haven't written up anything about it or published any code yet, but it's working pretty well and I haven't even had to fork Postgres with all the extensibility it offers!
My email is in my profile if anybody would like to chat about it
On writing some blog posts about things I’ve built lately (both at work and in my own time). Helps a lot with diving more deeply into topics than what’s reasonable for a “just needs to work” implementation.
Recently, a fairly detailed one on doing something semi-obscure with directory services on AWS. https://excessivelyadequate.com/posts/sadwsp.html
Love the domain name! I should start blogging as well. Seems like it would help in a similar way as teaching others.
Assertables: Rust macros like `assert!` for smarter testing, easier debugging, and faster refactoring.
https://github.com/sixarm/assertables-rust-crate
I have my side project: https://www.vidwiz.ai
Think "Cursor, for videos"
Very crowded space, but it's been fun making it!
I'm currently trying to implement an in-editor sandboxing/modding solution for the Godot game engine. It's hard work trying to make everything work the way people are used to having it, and even competing with GDScript.
https://github.com/libriscv/godot-sandbox
I originally started on it just to get into Godot.
Org-mode grammar (PEG parsing) for GNU Guile. Still at the very beginning, but hoping, that over time I can add more and more things, to make it a useful library for things like a minimal static blog based on org mode files or if any git hoster wants to have a good parser for org mode files to render readmes ...
That, and my personal website, using only HTML and CSS, and trying to keep it minimalistic, yet nice looking.
I'm working on an open-source event tracking infrastructure based on AWS (think Heap or Mixpanel but all infrastructure is in your AWS account and you own the data). It's incredible how much can be done just by combining AWS services.
https://github.com/manymetrics/manymetrics
A basic CRUD app around goal setting as a test for how AI tools can help write web apps to see how a modern web team could leverage this stuff to go faster, and maybe identify some missing pieces I could build one day. ChatGPT, Claude, Cursor, Copilot etc are genuinely great if you can already write code. They really let you blast straight through the mundane bits and focus on the hard stuff.
Working on an extension to my recently purchased outdoor AirGradient unit to add an atmospheric pressure sensor, a second temperature/humidity sensor, some connections to an external rain gauge and a way to power the unit without going through USB.
Many of the sensors and connections are small enough that I could have spun another PCB to replace the VOC/NOx module it comes with[1], but SGP41 ain't cheap & I wouldn't dare to desolder one from the existing module. So instead I'm going to try to use the extension I/O connector AG board has. Am currently waiting for my PCBs to arrive.
Speaking of PCBs. It is wonderful that it is possible to get 5 units of a prototype for a price of a coffee or two.
[1]: https://www.airgradient.com/shop/#!/SGP41-TVOC-NOx-Module/p/...
Shortest.com - AI writing my test suite for me, so I can focus on shipping features
I’m working on a generic way of getting human input within any automated workflow.
The forms can be dynamically generated within the workflow, and then call back with the response.
The docs still need some work and I plan on adding some video demos, but here it is so far.
https://humaninput.app
Working on getting communications back online in western NC. We are about to post a fundraiser to deploy 10 5G sites by the end of the week.
I’m working on marketing https://softwareengineeringhandbook.com/
We’ve experimented with various approaches to promotion, including HN, KDP, Amazon Ads, and most recently Reddit Ads. It's been interesting to see which strategies resonate with the audience, but we're still figuring out the best way to get it in front of the right people.
And marketing is really hard!
What is the unique value proposition of this book? How does it stand apart from the numerous amount of other books on the same topic?
Marketing is hard when there isn't a clear brand. Branding is hard when you don't have a very simple and clear differentiator to promote.
Thanks for the questions! I hear you it does sound like a generic book name. Well, we have the domain and we couldn't really name it to something else as we think software engineering has many elements and we wanted to cover them.
Our book isn't just a technical book on software development. Instead, it goes into the life aspects of being a software engineer such as migration and parenting.
Many of us have wished for mentors who could guide us beyond the technicalities, offering insights into personal growth and career navigation. Recognizing this gap, we've created a resource that provides practical wisdom.
By taking a holistic approach to software engineering, we address both personal and professional development in a way that few other books do. This unique blend sets our book apart, offering a clear differentiator that defines our brand.
Any resources you'd recommend to learn marketing (esp. comming from a software background)? Asking for software friend who's having trouble marketing his software business.
(Also good luck with your book!)
I found An Entire MBA in 1 Course really helpful. It actually goes through core business principles, covering everything from marketing and strategy to finance. https://www.udemy.com/course/an-entire-mba-in-1-courseaward-...
It looks like marketing within large companies is vastly different from marketing for smaller initiatives.
Working on a HomeAssistant plugin to control my home from my Macintosh SE/30. Compiling with retro68 and using my bluescsi for wifi.
Sounds really cool! Do you have a page describing your project with some pictures? It's funny I checked your post history before asking and found that you made the same comment I'm doing today to someone doing something similar one year ago! Is it what got you started?
Helping React devs move their state to the URL, in a type-safe way.
nuqs [1] started as a Next.js-only library, but recently I've been working on supporting all major React frameworks & routers (Remix, React Router, plain React with Vite etc).
[1] https://nuqs.47ng.com
iOS game to learn rhythm and drums [0]. It's MIDI based. Midi files to use as the tracks to practice on and Midi controllers to use as input. Here's a demo with my son on electronic drums [1]
Tech stack: Swift, UIKit, SpriteKit
[0] https://testflight.apple.com/join/Sy5573Uw
[1] https://youtu.be/RN2RRewR9B4?si=ic-_dmwp2sJGh94D
I'm working on a scripting interface for android that's based on js https://github.com/MurageKabui/PhoneDo
Basically a mobile app with an integrated IDE and terminal with custom commands tailored to execute js code that interfaces with native android features
I'm working on software that isn't source-available (as of now, anyway).
I've written an app that is aimed at a specific demographic (so I'm not linking to it), and I'm developing an improved backend admin app.
This involves mostly Swift, using UIKit, to produce an app that will run on iOS, iPadOS, and MacOS. The backend is PHP, and doesn't need much work.
Working on cleaning up my wood shop and trying to finishing my hand-tool wall.
Also building an easier way to add real-time metrics and monitoring to web applications: https://flexlogs.com
Also, this little side project for less overwhelming weekly goals: https://carpeweekem.com
A native MacOS menu bar app that gives you a deep breathing reminder every hour: https://github.com/p44v9n/deepbreath/releases/tag/v0.0.4
Functional but a few small bugs to iron out, then want to get a nicer welcome screen up and submit to the Apple App Store. Would love any feedback!
A toy compiler for analog quantum architectures.
Also, on my spare time, a tabletop role-playing game.
I’ve been building a copilot for an underserved language, and paused that in March with a little time since spent making a full language service: something where you can parse and resolve methods and types, and generally query for useful info. Perhaps it’s the root of a LSP server in future.
Building Serverless VPN, among the most recent work is an open source iOS app: https://UpVPN.app/ios
Same, same:
A custom Lisp: https://github.com/codr7/sharpl
A backend on top of Postgres: https://github.com/codr7/hostr
And a frontend in React: https://github.com/codr7/hostr-web
I'm working on an online drum machine for https://dopeloop.ai/beat-maker and the web version of a game called Asterogue https://asterogue.space
Working on an alternative to character.ai, they recently updated their UI and a lot of the fan favorite features are gone.
https://makebelieve.lol
Would suggest better AI model images on the homepage
Flux models look far better than these older stable diffusion ones
Working on an alternate reader and similarity search for hacker news: https://hn.garglet.com
Some features:
Search user profiles
Find similar comments
Find similar stories
Find similar users
See user karma next to their comments
browse comments in chronological order on stories
Recently got exicited about transforming my Tensorflow.js Parking mointoring application to a more general webapp that can do many things with Computer Vision, as well as upgrade from YOLO7.
https://github.com/oxedom/parker
Adding optional user sign-in to my word replacer browser extension: https://github.com/dan-lovelace/word-replacer-max. This is a precursor to adding generative replacement suggestions based on search terms. Even further, I'd like it to eventually analyze and de-trigger/disarm the copy on websites more broadly without having to define specific words or phrases.
Trying to finish this ultralight airplane with started to build with my father 14 years ago!
A real-time interface for my Fantasy Premier League… league. It calculates point totals up to the minute, which the official app lacks for some reason (can take hours for final point tallies).
It’s mostly an excuse to play with data processing with duckdb, remote APIs, and Pocketbase.
A tool that makes it easier to run Docker Compose projects on NixOS. It’s essentially a Compose backend that targets a mix of NixOS + systemd + Podman/Docker.
https://github.com/aksiksi/compose2nix
Working on https://aliveai.app/. It is a simple wrapper frontend around a complex ComfyUI workflow. There is a lot of different things you can generate with StableDiffusion but I wanted to make it as easy as possible to create photo-realistic images of people.
The App is mostly being used for generating NSFW images though (which is ok).
Building some prototypes of games with Godot. Mostly enjoying it.
A few gripes with the GDScript language though. Might switch back to C#.
https://godotengine.org/
I'm just dipping my toe into the Typescript water, with a rehype plugin that helps me turn markdown into <figure><img><figcaption></figcaption></figure> HTML.
The code's done, the yak shaving of packaging it as an npm module continues.
Avalonia version of my terminal Redis client https://github.com/cosmez/RedisMan
I've been doing a new Qt Widgets binding for Go - https://github.com/mappu/miqt
Prototyping a reactive UI library using DataScript[1] as the db and LWJGL as the rendering layer. I just want to see what happens.
1. https://github.com/tonsky/datascript
Rewriting my "UHF" app — a personal TV channel that plays video content I have on a hard drive to a schedule.
Also beginning to build a piece of furniture for "The Lab" (man-cave?).
https://www.videototextai.com/ - an AI transcription, translation, chat with your video/audio platform. We are very close to releasing an update where it is possible to caption any video in any language - perfect for making social media content.
I'm developing a scuba diving app for the Apple Watch Ultra. It's an algorithm and graphical user interface that provides critical real-time information to recreational scuba divers while scuba diving. (Nautosys.com)
Gnosi: https://www.gnosi.ai/ . Y You can auto-generate a constantly updated encyclopedia from a set of documents and conversations with an AI about those documents. Trying to make the process of having a Zettlkasten or personal notetaking system frictionless.
Looks promising. Do you have any success stories from using this took?
I've been recently working on a Minecraft-like sandbox game at https://github.com/Pannoniae/BlockGame.
The tech stack is .NET and OpenGL.
Progress has been a bit slower than I wanted mostly because I've been sick but we'll get to an MVP some day!
A browser based no frills TODO list for managing and sharing multiple lists without having to login or sign up: https://www.computedigit.com/list.html
Working on https://www.brief.news - a completely personalized daily newsletter on the topics you're interested in. We've just launched the ability to add custom topics, so you can create a newsletter on anything now!
This looks cool. Would be willing to pay for it if there was a RSS option.
torchtune (https://github.com/pytorch/torchtune) - a PyTorch library for fine-tuning LLMs, particularly for memory-constrained setups. Try it out and fine-tune Llama3.1 8B on a single RTX 4090!
Working on book summaries product - https://brieflane.com
Built a hardware device that sits between any computer and any printer and reads what is getting printed.
Primary use-case is to read receipt data from legacy POS systems without having to write software integrations.
Figuring out how to commercialise. Reach out if you have ideas!
fond memories. we did tv guides for a local cable system for a bit in the early 90s. they had a customer db, but no "export" beyond sending print jobs to their bigarse line printer.
but it was a parallel port and i had use of a luggable with a bidirectional parallel port so i'd haul that in once a month, hook it up, and have them run a "hello customer" fake billing run, which was hoovered up and stripped down to the address list we needed to mail to that month.
React Email Generator: https://reactemailgenerator.vercel.app
Just write a prompt and get the perfect email template for your use case.
I'm building android app for Teenage Engineering OP-1 (original) backups
Still working on https://evy.app/ - app for collectors, both professional and casual, to help them keep track of their items.
In short, it is an asset management system tailored specifically for collectors.
Working on my first foray into DML-based speakers. Low expectations, but more enjoying the fun of learning a new domain (and it’s a distraction from building garden furniture). Got various exciters, panel types, mount ideas ready, just taking the time out of the evenings.
Let us know how they sound.
I'm happy with boring old full-range.
Working on creating Java bindings for io_uring. Trying to get some better performance by batching downcalls and making the API Java friendly.
New for me, I guess:
- Doing a massive tech modernization for a global nonprofit.
-Ghostwriting educational email courses for agTech founders who want to convert more customers or investors.
- IF I'm good and get my chores done, I may let myself build a better way to apply for agTech grants.
Anyone know of a community where you genuinely try each other’s stuff and give feedback? Sometimes lose motivation aimlessly guessing what people want without having users to give actual feedback.
Kind of stopped writing code for a while. Been a few months.
Going back to the basics... a ToDo app
Working on DRAAS AI. Data analysis for geolocation compliance.
https://www.draas.ai
Just trying to get my personal site up and going: https://nassharaf.github.io/ideasthete/
Like the film Rashomon, had not heard of The Rashomon Effect.
I’m working on a replacement for file coreutils (touch, cp, mv, rm) which prompt before deleting by default and have a more sane api overall. Intended for interactive use on the CLI. In Rust.
Working on https://achromatic.dev - Next.js 15 SaaS boilerplate focused on web app functionality.
Just added MFA via authenticator apps + recovery codes today.
Finishing a web map widget suitable for geo-data visualization.
Building a narrative interactive novel/RPG thing. Wrapping music composition this week, writing about half done, game engine built, art finishing around Christmas.
Aiming to launch next summer, with various media.
A scavenger hunting app for couples. I have it working but I'm trying to figure out the gamification aspect.
Working on https://carbonimpacthq.com
We are building out the Vanta equivalent for sustainability and climate disclosures.
Working on a data exploration and visualization tool for SQLite. Some of the features include: ER diagram, browse table data, charting (bar, column, line, pie, scatter).
I'm currently working on a new blockchain technology called Roughchain. Earlier today, I shared the whitepaper in this HN thread [1], and I’ve already received some valuable feedback.
The core architecture is split into two components: a timestamping signing service and a P2P gossip network. By decoupling the gossip network, I'm simulating its performance using a Monte Carlo approach. With a basic gossip protocol, the simulation reaches ~10k TPS on a 100-node, randomly connected network (not fully connected), and I see a lot of potential for further protocol optimizations.
Initially, I considered a more resource-intensive approach using Shadow [2] for more realistic node simulations, as outlined in this discussion on libp2p's Gossipsub stress metrics [3]. However, the Monte Carlo method allows me to simulate the network more efficiently without needing to deploy full nodes.
In parallel, I'm exploring game-theoretical concepts for selecting signers and ensuring the system remains open to new entrants. One paper I'm currently diving into is "Collusion, Efficiency, and Dominant Strategies" [4].
[1] https://news.ycombinator.com/item?id=41687715
[2] https://shadow.github.io/
[3] https://discuss.libp2p.io/t/rough-stress-metrics-for-gossips...
[4] https://www.sciencedirect.com/science/article/pii/S089982561...
I'm working on this food logging app / personal health dashboard that I'm building for myself.
I was asked to log my food by a dietician I'm working with, because I was constantly feeling hunger after cycling training.
But all the solution out there were quite complex to input data, and I just want to write down plain text with some additional markup and be able to generate some graphs and recognize some patterns out it.
Decided to do it T3 and throw in all those weird technologies just to see what's out there.
- browser extension development framework from scratch
- doing a few SQL courses
- NAND to Tetris
- Graphical programming
Still researching trust. It's the deepest philosophical rabbit hole I've ever fallen down, but am now coming up for air.
Diverse opinions and even conflict in small groups can be productive provided there is trust. I’d love to understand trust better. Got anything to share or places to start?
I’ve been on a deep dive on the philosophy of harmony for a long time. Just submitted my second major article on “harmony of opposites” that deals with the role of conflict/tension in harmony.
> philosophy of harmony
Very interesting!
Alignment of interest and principles of conflict resolution and diplomacy are where I got to with "dynamic trust systems" at the moment. The whole project is to kinda push "beyond authentication". Yes I'd love to share some as I've been seeking proof-readers in some security communities. If you DM me via cybershow,uk (email in footer) we can chat.
from a someone'S slide from the 1970s:
"Knowledge of origin creates trust. Knowledge of capabilities creates trust."
I'd add "a mental model of someone's motives creates trust".
Both of these! And spot on with (mutual) motive analysis to come up with an "alignment" matrix.
I was inspired some time ago by Stella Rimington's writing (ex MI5 chief) that "identity" is actually a very poor basis for trust and authentication. It's unnerving watching the whole zero-trust show organise itself around notions of strong identity (as opposed to role and earned trust), which might turn out to be a rather silly thing to do.
I'm getting LLMs to modify your source code by writing code that modifies the Abstract Syntax Tree of your code rather than the code itself. I wrote a simple blog post to explain it: https://codeplusequalsai.com/static/blog/prompting_llms_to_m...
My initial goal is to let users make webapp prototypes and iterate on them by writing tickets for the AI to complete.
I for some reason call it Code+=AI: https://codeplusequalsai.com
I've just played with local models mostly, but I've found it difficult to get them to follow every part of the instruction.
For example they're oh-so keen on doing math themselves, even though they're shit at it and I instructed them not to do any math.
It's also hit and miss if they implement the right method or not, even with low temperature.
In my case I was experimenting with translating simple word problems into matlab scripts, so a resultcould be computed.
Do you find the AST approach helps? Or is it mostly just throwing compute at it, ie larger more better?
I'm still pretty early in terms of figuring out if this approach is better for larger projects. I can confidently say it works great on small things - for small changes on small files the AST approach works pretty well. You can say things like "add a click listener to the button that calls a function to tally the user's score" to a simple game and it will do it. That is, less than a minute after typing the prompt, you will see the code update in the editor and see your preview-webview update with your changes applied.
However, I have noticed that the AST code quality heavily depends on how common it is in the training set. I think I will have to add documentation to it through RAG or something - because OpenAI's models that I'm using seem to have limited experience writing esprima for JavaScript for example.
So it's hit or miss. In some cases I do feel like I'm throwing stupid compute at solving small problems and it's unnecessary - however, as I work on the project, it is getting better and better at successfully making the modifications. Some of that is me improving the prompts, some of it is OpenAI improving the models themselves, and some of it is the infrastructure I'm building for the project itself.
I did notice a huge improvement when o1-mini released. It is dramatically better at writing the AST code than GPT-4o or 4o-mini. I haven't tried Claude 3.5 yet but I've been hearing it does an exceptional job at code writing - not sure about my AST requirements though!
A little stable of websites.
www.fpgajobs.com
www.firmwarejobs.com
www.reportCardcomments.com
www.primeribcalculator.com
Trying to mimic mode 7 (Mario kart graphics) using canvas and JavaScript.
It is fun.
Taking my metaverse/game to alpha and crowdfunding.
Getting a job in cybersecurity again.
I just launched https://www.crowdwave.com a week ago, it spent 24 hours on the front page of HN.
Lots of people have visited but a launch on HN isn’t enough on its own. I’m trying to figure out how to get the word out to more people to kickstart it. The goal is for it to be a community that people return to as part of their daily online life. That’s not a programming problem so it’s hard (for me).
I'm working on an AI outlining tool in preparation for NaNoWriMo.
I'm building a hotel reservation system on tripoffice.com
a task runner/build tool thing, https://github.com/notnmeyer/tsk
I'm just starting a project to automate XBRL tagging
A scavenger hunt app for couples
I am working on (yet another?) DuckDB gui, with an emphasis on devops-y/back office workflows.
I want to join data from AWS with other sources and present a nice data-table UI, and perhaps allow taking some basic actions on a row, defining some filters, etc.
Have you ever tried to copy-paste data out of the AWS console? Truly a terrible experience.
Just finished a python package for a 'readability' calculator for Japanese: https://github.com/joshdavham/jreadability
Not the most impressive project, but hey, some of my friends found it cool!
On Mac, pbcopy and pbpaste don't work with images. I've written one that does in python, so I can right click on an image in Chrome, hit copy, then do pbipaste > foo.png and have it work. And then also do pbicopy < foo.png, and then be able to paste in Preview or whatever.
it's incomplete but https://github.com/fragmede/pasteboard-image
I'm in the middle of rewriting it in rust so it's easier to install.
Trying to figure out a bizarre performance drop in a merge sort I wrote for one of those daily coding problems.
https://bruceediger.com/posts/mergesort-investigation-1/
It seems that adding one node to a linked list causes a repeatable performance drop. That is, linked lists of say 2^21 nodes sort faster than lists of 2^21 + 1 nodes.
Trying to bring the finishing touches to a Common Lisp SSG I made during a handful of vacation afternoons (https://git.sr.ht/~q3cpma/make-website) and filling the resulting website (https://world-playground-deceit.net/) with more content. More motivated by #2 now that the generator does 99% of what I need.