Ask HN: What are you working on? (February 2025)

I'm creating an infinite canvas that has all your organization's code and documentation on it. If you zoom in, you can see the code, if you zoom out you see the big picture. By giving everything a place on the map, it becomes easier to figure out your way through the landscape and understand the systems. Different modes can you show you different things: code age, authorship (bus-factor, is the person still with the company etc), languages used, security issues. There's time-travel, think Gource for all software in your company, and maybe the most fun: a GeoGuessr for code. Select the repos for your team (or if you feel confident, of the entire org), you get a snippet and have to guess where it is. The plan is for LLMs + tree-sitter to analyze all the code and show relations to other systems, databases etc.

I had the idea 2 years ago, but starting building in earnest 2 months ago. Spending all my time on it now, minus 3 or 4 days per week of earning money. Currently looking for a GTM/sales-oriented cofounder in NL.

I live next to a school, so there's a low speed limit (30 km/h). Still, people drive like race drivers and the city hasn't ever responded to the residents' hopes of introducing a speed camera.

I wanted to have some data on how many people speed, the max speed recorded, that sort of thing. Things the city should be doing after many complaints of dangerous driving and people being almost killed on zebra crossings.

I have a doorbell camera, and by analysing the footage using OpenCV and some code, I can track how fast people drive if you see how fast they move between two known points.

Average speed: 46 km/h :(

Continued working with my team to grow my granddaughter who at 15 months handles a spoon and fork to feed herself at each meal; drinks from a cup without assistance; can clean her face and understand everything she is told or asked (though sometimes with a devious smile makes what an adult might consider a poor choice.... She is testing her boundaries like she is supposed to do). I have learned how to and produced 6 different embroidery patterns on various pieces of infant clothing. I combined multiple web based directions to create a Wi-Fi enabled USB (from a raspberry pi W 2) to enable a link from my computer to my embroidery machine. I made cookies and shared them with others creating a lot of joy I'm visiting with my grandson in another state, modeling good parenting and offering help where I can.

I'm working on a tool called Font of Web https://fontofweb.com that helps identify the fonts used on any website. It not only detects the fonts but also shows exactly where they're used (which HTML elements) and how they're styled (weight, line height, size, letter spacing).

My goal is to build a comprehensive database of font usage across the web. By collecting and analyzing this data, I believe we can uncover valuable trends, such as:

* Common font pairings * Popular heading fonts over time * Market share of commercial fonts * Top font foundries based on actual usage

I originally built a version of this four years ago and saw a surprising amount of organic interest. I've now rebuilt the tool from the ground up, switching from a Puppeteer-based crawler to an invisible iframe approach. (More details in another post)

Check out the current version at https://fontofweb.com. I would appreciate any feedback

I'm still working on Habitat. It's a free and open source, self-hosted social platform for local communities.

The plan is for it to be federated, but that's a while off yet.

I recently spoke with a Lemmy developer who gave me some advice on making it easy for anyone to host. I was struggling with the mess of supporting both docker and VM hosting. He told me that Lemmy uses ansible provisioning to install docker compose on the target VM so that the effort can be focused on docker support, so that's what I've been homing in on for the last few weeks.

- The idea: https://carlnewton.github.io/posts/location-based-social-net...

- A build update and plan: https://carlnewton.github.io/posts/building-habitat/

- The repository: https://github.com/carlnewton/habitat

- The project board: https://github.com/users/carlnewton/projects/2

I'm interested in uniform approximation with generalized polynomials -- these are linear combinations from families of parametrized continuous functions over some domain that satisfy some technical conditions, but its also fine to think of them as sums of regular monomials like 1, x, x^2, ..., x^N. This problem has been well understood for real intervals (classical case) for a long time, but I'm interested in this problem where we're approximating functions over complex domains.

There is a theoretically stable algorithm for the classical problem called the Remez exchange algorithm, and an extension to complex domains due to P.T.P. Tang in his 1987 PhD thesis at Berkeley. Theoretically Remez and its complex extension are very stable, but unfortunately implementations my advisor and I are aware of seem to struggle with large degree polynomials, where large is bigger than say n=45 -- errors begin to explode.

In any case, independently of this I've been learning more of the nitty gritty details of deep learning for a project at work (I'm a SWE in my day job, the math is more moonlighting), so to ground my efforts there I've been exploring deep learning approaches to this problem of complex uniform approximation, implementing results from various papers and tweaking things for my use case, and coming up with questions. That's much of what I'm thinking about this week!

Also, I'll be having a half-day long ADHD evaluation session on Friday -- so a bit apprehensive about that.

A C-based graphics engine/raycasting engine to make 90's games like Wolfenstein3D (1992) - but on a never-before seen scale.

The scale is RNG worlds like Minecraft. I've never seen that before with a Raycaster.

Here is my progress so far (I've had a month break)

https://github.com/con-dog/chunked-z-level-raycaster/blob/ma...

Not for profit, just for fun and exploration

I'm working on cataloging open source hardware designs.

When I'm starting a new hardware design, I find myself pulling up familiar boards (like Adafruit or Sparkfun's dev boards) as often as the chip's application note. I sometimes prefer a full reference project so I can get useful context like which voltage regulator they used or how the USB port is connected.

But, it's kind of an awkward process because I'll have to download the design files from Github and open it in the native CAD software (Eagle, for example).

I've been toying with how to solve this. I made a script to crawl Github for open hardware designs, then generate a schematic and interactive BOM for each design. Now, hopefully, you can search for "ESP32"[1] or "WiFi"[2] or "Bluetooth"[3] and get a number of designs to view in browser.

[1] https://www.openappnote.dev/tags/esp32/1 [2] https://www.openappnote.dev/tags/wifi/1 [3] https://www.openappnote.dev/tags/bluetooth/1

I recently made a little tool for people interested in running local LLMs to figure out if their hardware is able to run an LLM in GPU memory.

https://canirunthisllm.com/

I have celiacs, so I'm making a database of every single labeled gluten-free product in grocery stores. (I have 8,858 products across 243 grocery categories).

GF products are expensive and hit or miss, I really wanted something where I could keep track of my favorite items. I also want to let people rank them, so maybe I can discover the best gluten free hamburger bun (Rudi's Brioche), or beer (Glutenberg Blonde).

I'm also making a user submitted recipe section, so say you want to recreate a Taco Bell Doritos Locos Taco, it's easy to link to the products you need.

I'm not sure where this project is headed but I couldn't find any jobs working in this space so decided to make something to help myself.

I'm making a violin. I've been a programmer for just over 25 years, and recently decided I wanted to leave the profession. So I'm planning on attending the Violin Making School of America in the fall, and my project before then is to build a violin myself. There are some good in-depth video tutorials, as well as some famous books on the subject, and I'm planning to pair up with an amateur luthier in my town. The tonewood I purchased just arrived!

I'm trying to figure out how people do solo research, that means:

- Reading advice like https://eamag.me/2025/Good-Research

- Figuring out problems like https://openreview-copilot.eamag.me/

- Doing analysis like https://eamag.me/2024/Automated-Paper-Classification

- Figuring out where scientists are located like https://eamag.me/2024/ICML-2024-on-a-map

I'm tinkering with the USB HID specification for power devices for all the wrong reasons.

Historically, UPSes had various proprietary communication protocols over serial ports. Nowadays they usually have a USB port, but I have a bunch of very old APC UPSes with just the serial port and/or the expansion card slot (which is usually just another serial link plus power on a edge connector).

A normal person would just use NUT or apcupsd over serial and call it a day. A bored person would write a USB HID power device stack and serial protocol acquisitors to give these UPSes USB ports. An insane person would add projectors from the USB HID power device stack to serial protocols so that they could use whatever communication card they want on any UPS they have (for example, a CyberPower RM205 card plugged into an APC Smart-UPS).

Why? Because apparently I'm insane and I needed a break from working on delinking executables back into object files, another heretical project I've worked on for the past couple years.

I've just started and I don't know if I'll finish that, but it's something I need to work on to exorcise that particular nagging thought out of my head.

I'm a long-time FreeCAD user, and one of my annoyances is that long-running operations lock up the entire UI and can't be aborted. This is particularly annoying if you realise you made a mistake, but have no way to go back and correct it without waiting for the operation to complete first. Or you kill FreeCAD but then you don't get to save your work.

So for my first contribution to FreeCAD I'm working on fixing this.

The underlying CAD operations are done by "OpenCascade", and at first I thought OpenCascade had no support for aborting operations part of the way through. So my first implementation was to move the operation into a child process and give the user a dialog box that would allow terminating the child process.

But it actually turns out OpenCascade does support aborting the operations! So now I'm working on doing it the OpenCascade way.

My PR is here: https://github.com/FreeCAD/FreeCAD/pull/19796

A small library which uses a fork of OpenSCAD which adds Python support: https://pythonscad.org/ to allow writing out DXF files and G-code and modeling how G-code will cut in 3D:

https://github.com/WillAdams/gcodepreview

Writing it in a TeX editor using a Literate Programming system developed in the course of working on it, the PDF should give both a good overview, and provide all the code.

It has greatly expanded/restored my math/geometry and has me looking forward to how to implement Bézier Curves and surfaces using similar math (since G-code and CGAL are fundamentally limited to lines and arcs and what can be easily made from such constructs).

I’ve deleted all my social media apps (including YOU - LinkedIn).

I’m trying to really see and feel what’s actually missing in my life and trying to build it. Right now I just want to see what my friends are up to in a non-curated way.

Very cliched, but I am working on my own Lisp dialect. I want a more streamlined syntax and keywords, but keep the "batteries included" idea from CL. So, Scheme/Clojure syntax with a CL live image approach, including condition systems, and a fleshed out standard library.

Initially was aiming to use MLIR or at least LLVM but will probably try to handroll to a) reduce dependencies and b) as a learning experience.

The bootstrapped is written in CL with no dependencies and hopefully soon it will be self-hosted.

In between figuring out what to do after a decade of work on Micro (https://github.com/micro), I started a new project called Reminder which tries to provide a single clean app and API for the Quran, Hadith and names of Allah. Maybe some of you would find it beneficial. It tries to put English first since most of us are non Arabic speaking and cultural from the west.

https://reminder.dev

I’m finally trying to jump on the AI train for a long ride instead of getting off at the first stop.

I’m currently creating a new fan site for Marvel Rivals (https://marvelrivals.app), and I’m trying to introduce new types of features using predictive analysis, and further, use some DL maybe to understand specific player behavior and do stuff like find cheaters. I am failing so far.

I thought it’d be easier to throw data at the magic AI monster, but it’s still garbage in -> garbage out. It makes me respect AI engineers a lot more.

I wish there was an easier way to apply AI to this kind of stuff, on the how to do better data analysis. Ideally, I’d hook up some tool to my Postgres db, which has a couple tables but everything is named appropriately and has references. Then the tool would output correlations, patterns, stuff people would find useful and interesting. Instead, right now, I think I have to make those guesses and then build models that will either support my hypothesis or reject it, but I don’t know ahead of time and it relies on my gut feelings.

I've just published my first novel for adults, The Dark Sorcerer's Intern, my bid to bring back fun and comedy to a fantasy genre that has spent years in a grimdark rut.

The relevance to hackers is that unlike most fantasy where spells are cast with hand motions, magic words, or spell ingredients, there's actually an explanation for why that works and makes sense.

https://www.amazon.com/Dark-Sorcerers-Intern-Humorous-Fantas...

I'm working on a website that anybody can update by calling into the phone number in the url. https://715-999-7483.com/

Working on "wanting to live". It's hard to create desire within oneself when one has experienced intense sorrow.

Been trying the "do the thing, and desire comes after" for many things (baking, piano, skating, ..), but that hasn't really worked. What has seemed to work is connecting with people (crucial that they know how to connect back).

Made a little web app that helped me communicate: https://azriel.im/tears/

(I could just point to the number when I couldn't talk/listen)

I've working on Colanode, an open-source & local-first Slack and Notion alternative that you can self host. You can use Colanode for real-time chat, as a knowledge center, project management or file storage. As a local-first application, Colanode offers full offline support, allowing you to work even when you’re not connected to the internet or the server is not available. You can host it in any environment (with minimal dependencies), giving you full ownership and control over your data.

https://github.com/colanode/colanode

A tool for building WCAG accessible Tailwind-like color palettes for UI/web design. :)

https://www.inclusivecolors.com/

> Instead of only working with a handful of colors, you can create a whole palette of swatches at the same time so you can see if they look good together.

> Precise control of every shades/tints in each swatch rather than being limited by autogenerated colors.

> See which color pairs contrast as you edit so you can create a palette with built-in WCAG accessibility. This way you can plan in advance which foreground colors (for headings, body text, form fields and so on) should contrast on which background colors, so you can avoid running into surprise low contrast issues later when designing.

I'm building an interactive, web-based Python tutorial site intended to help with learning basic syntax. Originally it was for my kids who wanted to learn to code, but... might be useful to others.

https://learnpy.dev

The content needs some work, but I'm pretty happy with the framework / UX. I would love to get any feedback from folks who check it out!

(The first section is just multi-guess questions as part of the introductory content. Try any other section to get the full in-browser-code-execution experience, which uses client-side Pyodide under the hood.)

I’ve been improving the developer experience of the extremely janky Java Spring app that powers the most popular open source real time transit app, OneBusAway.

Last month I added Dockerfiles and a docker-compose.yml file to the project to make building and locally running it a breeze. Earlier today, I finally had a chance to figure out and document how to debug the app, which should greatly improve quality of life for anyone trying to fix bugs or add features to the backend. https://github.com/OneBusAway/onebusaway-application-modules...

I'm developing ParaNoia, an application that allows people to securely share passwords between each other (https://github.com/jonesconnor/paranoia).

At my organization, we have a corporate-approved password manager, but shared folders are disabled. If someone needs a FID password, or something of the like, there are people that send the password in plain text over Slack, Teams, etc. and not everyone deletes their message containing the password afterwards.

You enter your secret on the website and it gets encrypted on the client-side before being sent to the database. A one-time-use URL is generated and then you send that URL to whomever needs the password. The backend never sees the secret in plain-text.

The tool is meant to be self-hosted - but I'm in the process of deploying it publicly for people to try out. I got the idea from my brother -- he was doing an internship at Tesla and said they had this being used internally, so I figured I would create my own as a little side project that I can then implement at work.

I've been updating my HN bot (watch comments for keywords and post to Slack/Discord) written in Crystal to use raw SQL instead of unmaintained ORM.

Turns out the whole app needs only ~ 10 SQL requests, and it's way funier to write modern SQL than fighting the ORM.

The new code looks like this :

    db_message = Model.save_new_message!(@conf.db, DbButler::Hn, item, DbState::Processing)

Than I have a Model module with all the interactions with the DB

    def self.save_new_message!(db, butler, external_id, state) : Message
        sql = {{ read_file "./db/save_new_message.sql" }}
        db.query_one(sql, butler.to_s.downcase, external_id, state.to_s.downcase, as: Message)
   end

(thanks to Crystal ability to read a file at compile time - I can write raw SQL in a file with syntax highlithing and maybe typesafe if I connect the DB to the editor)

The land page is not ready, but the bot has been working for me for months https://newsbutler.xyz/

I taught my parents how to use LLM chat apps. I was pleasantly surprised to see them use it all the time. And even more shocked to see them pasting entire whatsap messages containing passwords, upload income tax files, and a lot more private details with LLMs. They rarely pause to think about privacy/security before sharing info with LLM services. So I'm working on an interface that works as a privacy filter, making sure the private info does not leave the device. It redacts /anonymizes/obfuscates private information from what we share with LLMs via on-device model, and plugs back the output with the private info to make it appear almost similar to the output as before.

I've recently been prototyping a mobile application to track your food nutrition. The key feature lies in auto-detecting the food based on a given image, and breaking it down into it's ingredients and then into it's macros.

Existing apps such as MyFitnessPal and HealthifyMe fall into two ends of the spectrum where you either need to add ingredients one by one, or your food is logged with a standard macro count where you cannot change the ingredients used.

Weit ideally provides a seamless experience in taking a picture to retrieving ingredients to retrieving macros per ingredient. Once that's sorted, food tracking should be granular enough to build intelligence around it to improve one's diet based on their requirement.

Honestly, I used to constantly struggle with the realisation that none of my ideas are unique and whenever I see someone having built something similar, I feel like I'm wasting my time. I'm getting better at dealing with it now though.

I've been building https://canine.sh for the past year, which is an open source Render / Fly / Heroku, etc.

It's based on some learnings I've had in the past building where building on managed platforms like Heroku and Render, and watched our costs explode, with an annoying amount of vendor lockin.

It uses Kubernetes under the hood (which you can now get fully managed for $12 / month on linode), which lets you take advantage of a ton of things that Kubernetes does really well, like automatic healthchecks, zero downtime deployments, auto scaling, etc, while also making it easy to use for solo developers or small teams.

The additional benefit of Kubernetes is that it's also possible to host a bunch of other stuff in your cluster via Helm charts, that you’d normally have to pay for like: Sentry, Wordpress, Postgres, etc.

I'm a fashion nerd. I've been working on an outfit tracking app for the last ~4 years (but only really pushed hard on it the last 3 months). I found I kept buying clothes that I never ended up wearing. Either because they didn't fit with what I had, I already had something like it, or it simply wasn't my actual style (although I thought it was). So, I built my own with the simple goal of buying less clothes, and throwing away fewer clothes.

There are plenty of apps that do outfit tracking, with some basic stats. But they all have a few or more of these shortcomings (from my perspective); unpleasant UI, no cross device syncing, lack of detailed usage statistics (e.g. cost spread over time by garment category), some categories just not supported, pushing a specific lifestyle such as Capsule Closets, or just plain focused on recommending what to wear using some mediocre algorithm that doesn't understand cuts and how different pieces fit together; basically only suitable for capsule collections.

These apps all have a lot of downsides too in common, which I haven't been able to solve either yet; ultimately you must start with an inventory of your clothes, and then work from there. It takes ages to catalog and import your clothes, and I haven't found many existing product that lets you export if you've even done it before. And on top of that, you have to be quite rigorous at tracking what you wear; the more data you have the more insight you can get from your choices.

I finally published on iOS a couple of months ago. No traction, and I don't expect there to be. I won't argue that my offering is better than any of the competition, but I've tried most of them (and wasted colossal amounts of time onboarding onto them) and found none fit my need properly. It's still very much work in progress, but I find myself reaching for it multiple times per week to inform my purchasing habits.

https://procloset.app

I'm working on Somleng (https://github.com/somleng) - an Open Source alternative to Twilio, that's being used to save lives by powering the National Early Warning Systems of Cambodia and Laos.

We're also building OpenEWS (https://github.com/somleng/open-ews) which will be the world's first Open Source Early Warning System Dissemination Platform.

We're a small 2 person team trying to make a big difference.

Firefly is a typed full stack programming language:

- object capabilities

- implicit async/await

- immutable collections

https://www.firefly-lang.org/

It's small and (hopefully) fun, and quite usable already. If you try it out, please share your thoughts!

I’ve always wanted a large wall mountable e-ink display that I could update periodically - and it didn’t need to be plugged in (or drill holes in a wall to hide the power cable).

The displays are really expensive so I’m looking at taking 12 kindles apart and mounting them in a 3x4 grid. They cannot seamlessly touch at the edges so I’m looking to include that as part of the larger aesthetic then ignore it.

I’ve figured out a few possible approaches and the software/service side - next step is to order 10 more kindles and get to work.

A new YouTube app/player, for my kids.

It allows us to control the algorithm. It’s all LLM translating to YouTube search queries under the hood.

Visually it looks the same.

The suggested videos come from predefined buckets on topics they love.

E.g. 33% fun math, 33% DIY engineering, 33% creative activities.

Video recommendations that have a banned word in the title/desc don't get displayed e.g. MrBeast, anything with Minecraft in it, never gets surfaced.

For anyone interested in using it, send me an email.

I'll put you on my list. And you can contribute ideas to our community Google Doc.

jim.jones1@gmail.com

I have been working full-time for about 15 months on a product to store real-world entity-relations in a graph (using AI/ML for extraction). The idea is to extract entities and relations deterministically from text (using AI/ML for clues about type and position of entities in text).

It is very much a work in progress with lots of commented out code which are just experiments.

https://github.com/pixlie/PixlieAI

A game: you're stuck on a 1970s spacecraft and you have to program your way home using 6502 assembly. I wanted to learn more about old 8 bit cpus so figured I would make a game out of it

After work hours I'm continuing work on my Saas for hairdressers. There's some big players there but I feel like I can at least still try.

I'm honestly really surprised about how much I get stuck on business logic decisions. I went into this thinking making appointments, basic managing of employees and all that would be simple and relatively similar across salons.

Additionally I'm considering where I should move to. I wish to live in a place where owning substantial land for homesteading in a relatively climate safe area (relatively doing a lot of work there but imagine not already arid or with high storm risk) is not completely out of grasp. My region of Belgium is too densely populated for this. Even if I'm not moving to a different country even next year I figured it's the kind of thing that takes a stupid amount of preplanning.

I've been working on an iOS app that aggregates cinema showtimes across chains and independent theaters in the UK.

I moved to London some years back, and was pleasantly surprised by the vibrant cinema scene, that seems to be in a steep decline in so many places. On any day of the week, one can find independent films, old and new classics, Q&A's with filmmakers etc. playing in one of the many theaters across the city. Staying on top of it all is a chore though, and I found myself missing out on screenings regularly, because I didn't check that one cinema's website on time.

This is also my first time building and releasing an independent app. The journey from research, backend development and learning SwiftUI has been a trip. Released on TestFlight a couple of weeks ago.

https://qino.app