alt Hacker NewsAsk HN: What Are You Working On? (June 2025)
What are you working on? Any new ideas which you're thinking about?
Comments
Getting close to dark-launching a web-app that scans a domain for common issues - there's a heap of "performance measurement", "security analysis", "SEO scan" sites - but nothing that really does them all - a one-stop-shop if you will.
It's been a fun ride co-coding with Claude (Sonnet 3.5 > 3.7 > Code). Already it found a bunch of interesting bugs on a heap of my own sites, older employer sites, friend's sites.
Started as a simple Django web app, extended to Celery+Redis, now also leveraging CF Workers and R2 storage.
Was bourne out of an observation that some sites I have been working on missed some crucial things like domain expiration of asset-domains, mis-configured CORS or SSL certificates, http header and meta collisions, missing/wrong redirects for http/https/www/no-www etc.
I've built a Reddit-like community platform in Go. Users can create their own sub-communities, and within them, set up different categories and boards. Posts can be voted on, and board types can include regular posts, Q&A, or live chat. It's like a hybrid of Reddit and Discord but leans more towards a traditional web community. It also supports server-side rendering, making it SEO-friendly. This project is an extension of my previous Hacker News clone, dizkaz (https://news.ycombinator.com/item?id=43885998). I'm currently working on implementing submission rate limiting and content moderation, which is a bit challenging, but it should be ready for launch soon.
currently in the process of building a small CPG/food lab/accelerator/incubator. I have access to talent, supply chain, logistics network, retail space and a little bit of capital. I want to take and implement the VC model for SaaS startups in Mexico. I'll have a very specific selection process for products we would take on and I'll provide all the infrastructure to get the product onto store shelves in Mexico, USA and hopefully Europe.
I have no idea what I'm doing but I never really have. I've always just figured it out on the way.
I'm tinkering with a little DSL for declaring embedded in-memory databases, to be emitted as C++ code. I've noticed that I keep creating assemblages of STL containers which feel like tiny databases, optimized for specific queries and constraints: a couple of vectors with structs, some maps or sets, maybe some running totals or min/max references, all wrapped up in an "engine" object governing whatever process it is that uses the data. Wouldn't it be nice if I could declare such a thing, SQL-style, and let the machine work out the details?
There's a little inspiration from the MLIR ecosystem here, which makes heavy use of `.td` files for code generation. I want to write a schema file, defining some tables and indexes along with the queries and procedures which will operate on them, then have this compiled to a C++ header file I can include, where the schema is a class and the queries/procs are methods.
I have no idea how far I'll get with this, but it's always fun messing about with weird little languages, and I'd like to see what programming in this style would feel like.
I’ve been building my wife a budget tracking dashboard for reporting on PPC ad campaigns.
At any given time, she’s working with any number of clients (directly or subcontracted, solo or as part of a team) who each have multiple, simultaneous marketing campaigns across any number of channels (google/meta/yelp/etc), each of which is running with different parameters. She spends a good amount of time simply aggregating data in spreadsheets for herself and for her clients.
Surprisingly we haven’t been able to find an existing service that fits her needs, so here I am.
It’s been fun for me to branch out a bit with my technology selections, focusing more on learning new things I want to learn over what would otherwise be the most practical (within reason) or familiar.
My background is in NLP, research, and startups. I joined a power company where I saw a clear opportunity to use AI for automating equipment inspection from drone images.
But the environment made it hard to move fast. The systems were outdated, and there wasn’t much support for building AI tools in-house. That experience made me realize I needed to grow beyond the modeling layer. There were things I wanted to build, but I didn’t yet have the full skill set to do it on my own.
So I’ve been learning full stack development. I had built a small chatbot app before, but this time I’m applying what I’m learning toward a focused MVP for the inspection work. It’s been a practical way to connect what I know with what I want to make real.
Wut.Dev (https://wut.dev) - a fast, client-side, privacy-focused, alternative to the AWS console.
I got tired of using the AWS console for simple tasks, like looking up resource details, so I built a fast, privacy-focused, no-signup-required, read-only, multi-region, auto-paginating alternative using the client-side AWS JavaScript SDKs where every page has a consistent UI/UX and resources are displayed as a searchable, filterable table with one-click CSV exports. You can try a demo here[1]
[1] https://app.wut.dev/?service=acm&type=certificates&demo=true
https://github.com/stravu/crystal
Crystal is a re-imagining of what an IDE means when AI drives development. Traditional IDEs are designed for deep focus on one task at a time, but that falls apart when you have to wait 10-20 minutes for an agentic task to run. Crystal lets you manage multiple instances of Claude Code so you can inspect/test the results of one while waiting for the others to finish.
A free meeting-scheduling tool with all the pro-sounding features without any restrictions or commissions. Think Calendly, but free and much better.
I was hoping to make a piano practice assistant for my kids, that would take sheet music in MusicXML format, listen to the microphone stream, and check for things they frequently miss like rests, dynamics, consistent tempos.
Surprisingly the blocker has been identifying notes from the microphone input. I assumed that'd have been a long-solved problem; just do an FFT and find the peaks of the spectrogram? But apparently that doesn't work well when there's harmonics and reverb and such, and you have to use AI models (google and spotify have some) to do it. And so far it still seems to fail if there are more than three notes played simultaneously.
Now I'm baffled how song identification can work, if even identifying notes is so unreliable! Maybe I'm doing something wrong.
This goes against the grain - but I’m starting to reimagine enterprise & manufacturing workflows using AI.
Very good leadership support, small (but great) team, huge mandate and uncertainties - and it’s quite exciting.
We are also hiring - AI+Full Stack https://cenovus.wd3.myworkdayjobs.com/en-US/careers/job/Seni...
I am working on building Flexprice(https://flexprice.io/), an open source monetization platform for AI and Agentic companies.
This week, we’re doing a 5-day launch week, where we’re shipping a new set of billing features every day. Github link: https://github.com/flexprice/flexprice
JSON API to integrate with QuickBooks Enterprise / QuickBooks Desktop https://qubesync.com
Spent 14 years slogging through a custom implementation with my previous company, and didn't want my pain and suffering to go to waste. Just spent a few hours yesterday to replace that app's integration with my new api and got a pretty good diff:
117 files changed, 258 insertions(+), 10032 deletions(-)
Building some software for the Trim UI Brick. It's such a neat little thing; being able to turn it into an offline-first device that can handle my data needs would sincerely fill me with joy.
Right now I'm making an Anki-style flashcard system in Rust (https://github.com/jmahmood/Cardbrick/). It's amazing that all of the annoying stoppers I've had with build systems were blown away by Gemini. It's also nice to actually build something that is not another web application.
(I say this knowing there is zero value in me posting this in a thread with 1134 comments already)
It occurs to me that I need a separate electronics workbench, more than the 2 square feet that I can make by moving my keyboard out of the way. So, this month... my goal is to have everything set up on a workbench.
Generating an estimated $130 million per day (100 Megatokens/second) worth of GPT4 tokens at home will have to wait (plus I'd need to upgrade the power and AC in this room a bit to handle the estimated 750 Kw of power it would take)
Recently, a lot of reverse engineering. I’ve been writing them up on my blog, so there’s a growing list of technical deep dives from this year: around Letterboxd with mitmproxy[0], iOS Shortcuts deeplinks[1], the QR codes for Fitness SF[2], and binary patching some non-open source code[3]. Hopefully followable even if you don’t have as much debugging or reverse engineering experience.
[0]: https://blog.alexbeals.com/posts/extracting-letterboxd-token...
[1]: https://blog.alexbeals.com/posts/reverse-engineering-ios-dee...
[2]: https://blog.alexbeals.com/posts/debugging-fitness-sf-qr
[3]: https://blog.alexbeals.com/posts/start-process-extensions
A screen reader for linux. My aim is to carry around my Raspberry Pi 500 or some other mini keyboard with a tiny computer embedded in it and have it serve as a fully functioning computer.
My hope is to make it easier to use a computer blind than with my usual workflow with a monitor.
Just writing posts for my blog on personal experiences with startups https://developerwithacat.com . Am taking a break from any serious building, bit tired of failing. Using the blog as a form of self therapy.
A TV-TV video calling system (raspberry PI + off shelf camera and mic). Idea is have a more immersive experience where you can talk to entire family instead of just a persons face (or parts of face in the hands of my parents or kid). Next stage is to add games to the calls. Have some games already prototyped with good feedback from kids.
Still thinking of how, what and when to open source.
Working on a silly side project called SinkedIn — a parody of LinkedIn but just for posting failures, screwups, and embarrassing moments. Staging is live here: https://sinkedin-staging.vercel.app/ and GitHub repo is: https://github.com/Preet-Sojitra/sinkedin. Pushing to production soon. UI is rough, I’m not a frontend person — bear with me! All sorts of contributions are welcome.
Two person Micro-SaaS which helps employers collect one way video interview screening responses from candidates at scale
Here's Hirevire’s #buildinpublic stats for June 2025!
MRR Metrics
$4263 MRR (+18.79% MoM ▲)
$348 is the average lifetime value and ARPU is $51.35
14.71% Gross MRR churn rate and 13.43% customer churn
$119,426 total revenue (last 12 months)
2 years and 10 months since launch
8.1K unique visitors, 4K from Google organic
$14,258 one time revenue collected
29634 (45% MOM ▲) applications collected
Conversion numbers
3.18% Visits to Trial signups
8.46% Trial to paid plans
Among many other things, I'm formatting many of my recipes and working on generating LaTeX to make a physical recipe book.
My son has inherited my love of cooking and baking, so we'll refine the book, add comments and photos, and eventually print and bind copies for our family and friends.
I also am hoping to laser engrave some old cookie sheets with one of my grandma's hand-written recipes. The problem I have is that it's rather faded, and I don't know yet how to make it pop for a good contrast.
Oh !
An ISC-licensed implementation of several Content-Defined Chunking algorithms in Golang at https://github.com/PlakarKorp/go-cdc-chunkers
Whenever you have redundant data you want to store / transfer, this library lets you perform fast content defined chunking
Partly for my own use but largely for learning purposes, I developed a multi-platform note-taking app using Flutter and Dart. It's a bare-bones mobile-first app that stores notes as *.md files to support SyncThing-based synchronization. The UI is inspired by Android apps like Omni Notes, Material Notes, Noteless, etc., which however are either not multi-platform or not aimed at SyncThing-based synchronization. Obsidian's mobile app, in my opinion, is not as mobile-friendly as these apps. Joplin is the closest I could find but which I did not like.
It's tested only on Android 10 and Windows 11. Bring done with Flutter, it should work on iPhone, Mac and Linux too but would need building, testing and fixing various issues found.
Had I known this would take me 3.5 weeks (dedicated time) and 6100 lines of code (including comments), I would not have done it. Ideal would have been just a week.
Currently closed source.
I'm working on a git(1) implementation written in Python. I'm following along the amazing James Coglan's book. Currently 23 chapters in, about to finish off the Part II of the book. Can't wait for the final, third part.
I love books like this one. Some other examples are "Crafting interpreters", "Writing a C compiler", "Building a debugger", and a couple other lower quality ones. The potential in this space for aspiring technical writers is enormous. Let me know if you know some other books that guide you through implementing complex systems software from scratch.
I am building https://balancing.services to improve transparency of European balancing markets.
The balancing markets are used to keep the power grid in good shape, by smoothing out any last minute mismatches between energy production and consumption.
The project started out of frustration of not being able to get this information without friction.
I wanted to learn a bit about backend development, so I've been building my own version of soundcloud with supabase. Main thing I've learnt so far, auth is flipping complicated. But it's been really fun! The audio compression is done clientside with ffmpeg and WASM, I'm pretty pleased with that approach. Everything is pretty busted atm, but I'm trying to get to a 'walking skeleton' then polish. I've been devlogging the process as I go for fun.
I'm currently developing a middleware that connects Nvidia PhysX to GameMaker. There's still a lot of work left but I have most features working in some capacity. Dynamic and static actors, primitive/convex/triangulated shapes, joints, character controllers, GPU accelerated PBD particles and deformables, etc. GameMaker is primarily a 2D engine and offers limited options for 3D, but it is possible if you know how to use vertex buffers. I'll probably post it here once it's a little farther along, but I'm pretty proud of my progress so far. I'm hoping I can use it to support myself in some way, but there's a lot of anxiety in selling a niche project like this.
https://thecoderssage.com Trying to build a platform for creating specific tailored coding lessons from prompts or your own code, with interactive coding, quizzes and exercises.Kind of youtube shorts but for coding lessons, on demand.
I'm working on Fro (https://fro.app)
Haven't released properly yet - not sure if it's stable but oh well.
I don't like using my personal email to sign up for things. But there are definitely things that I do want to sign up for - newsletters, try out some services.
I know there are temporary email services, but I actually want to use these services. Of course there is Apple email that forwards to your real email.
But, I also don't want to flood my inbox.
Anyway, I wanted to receive these transactional emails in my personal Slack.
So, that's what Fro is for (https://fro.app)
- Sign up - get an email address - link to your Slack channel
And you can now catch up on those newsletters via Slack.
On a whim, I bought a pack of playing cards at the supermarket. Now I'm learning how to play card games.
The card maker has its own web site with the rules for playing all kinds of card games, and it's filterable by number of players, including many games for one person.
I'm building Talo (https://trytalo.com) - an open source backend for Godot and Unity games with leaderboards, analytics and authentication.
I've been working on making it easy to drop in socket-based multiplayer with "channels". Players can join channels and they can share messages, state updates or notifications over a socket connection. You can use it for chat rooms, lobbies/matchmaking or async multiplayer.
One recent addition is "channel storage": a shared space for players to read/write/update/delete data. This opens up saving and loading shared worlds between players in just a few lines of code.
Everything is open source, including the frontend dashboard, backend, Godot plugin and Unity package. GitHub here: https://github.com/TaloDev.
A web-based level design editor for game devs, Rust-pilled.
https://childrensbookforall.org
My wife (who is a psychotherapist) started this and I am helping her with this. We are using specially curated children's books as a medium to talk about social, emotional and psychological aspects of mental health among adults, adolescents and teachers. We are also building communities and support groupsaround children's books.
This is in India where talking about mental health can often be a taboo subject. People who need/want to talk about this also find it hard to express and there are limited spaces which give you opportunities to do so. We found the abstractness of children's books as a great evocative medium. They also promote play, wonder and joy - aspects which positively impact mental health of individuals.
The project started with a personal journey of grief my wife experienced (death of a parent, diagnosis of other parent with Stage 4 cancer).
https://github.com/thiagoharry/weaver-interface-metafont
I started by trying to reimplement the METAFONT language, adding support for real-time rendering with OpenGL. Eventually, I decided to introduce some incompatible changes, creating a new language. But it still retains a syntax and internal logic very similar to METAFONT.
This new language also supports animation, and since it is part of a larger project (a game engine), it can be used not only for font rendering but also to generate textures and sprites for games.
The language is successfully compiling to WebAssembly, and I’m currently working on a web page with tutorials, documentation and examples where you can modify the code and instantly see the results. Since this is a literate programming project, there is also an English and Portuguese version of the code. But the english version still needs considerable polishing.
Solo building this project: https://www.usesubwise.com/
A side project that started out of personal frustration: I kept getting charged for subscriptions I forgot I even signed up for.
Subwise helps you track all your recurring payments across platforms — with smart reminders, spending insights, and cost optimization tips (like switching to annual billing or downgrading tiers you don’t use). Think of it as Mint meets a subscription-focused dashboard, with a browser extension that auto-detects subscriptions during checkout.
The bigger picture: I want to reduce "subscription fatigue" and help users stay financially aware, especially as more services go SaaS/paywalled. Eventually thinking about small biz or team-level use cases too.
Tech-wise: Built with Next.js + TypeScript, Postgres + Drizzle ORM, and a lightweight Chrome extension. Stripe is integrated for premium features. Everything’s still MVP, but slowly improving!
Would love feedback if anyone here manages a bunch of subscriptions (or forgets them like I do).
Working on FilePond v5.
Entering year three of a complete rewrite. It’s kind of ridiculous but as I’m still enjoying the process of trying to built/craft a performant and flexible file upload web component I just keep going.
V4 is live on https://filepond.com, plan to release v5 before the end of summer.
Website where you can download lists of all registered domain names for each domain zone in the world.
Currently available: 274 million domain names across 1570 domain zones.
Domain lists are updated daily.
Download via website or via HTTP REST API.
Can be used for parsing, marketing, automation, research and whatever else.
Phoenix — https://github.com/shofetim/phoenix
A multi-server process supervisor. Existing init processes (systemd, runit, s6, etc) work great on a single server but when you need to manage/deploy many servers, the tooling gets really complicated (K8s). Phoenix extends the process supervision model from one server, to many. Run this thing once / keep one copy of this around / keep this running on all machines that match pattern X etc.
Turns out the (obvious in hindsight?) problem is automated but simple networking. Currently digging deep into wireguard based overlay networking before rolling the next version of Phoenix out.
I've been off work for two weeks to recover from surgery, and have been playing with a couple projects throughout the day between rest and physical therapy:
- A home-rolled router/firewall: Using yocto to create a distribution for a router/firewall for my home network. It started as an exercise in wanting to have more control over the security of my home network, as well as see how nice of a UI/UX I can tease out of an LLM. It's also part of a (seemingly never ending) consolidation of homelab services.
- A SNES Reverse Engineering setup: A nephew of mine is starting to get into video games and is starting with a SNES but his system broke. I'm working on helping repair the console, but am also trying to set up an effective "LLM + Ghidra + SNES emulator + image generation AI + asperite plugin" to allow him to swap sprites and text in games to add some creativity and learning to the experience.
- A personal assistant system: Experimenting with agents to create a personal assistant for our house, and seeing to what extent the agents can be helpful and how much hardware is required to run something like that in-house.
- aztui: A TUI for exploring and interacting with Azure resources. I'd like to add some caching/pre-fetching logic to make the interaction with the interface snappier (one of the main motivators to create it).
I've been using GPT pretty heavily throughout, and it has been a lot of fun both using it, and spending some dedicated time looking at the models themselves along with the frameworks that support running and integrating them.
Finally cleaned up a free online Python course: https://proficientpython.com/
I wrote the articles/exercises/projects a few years ago, but now I've made interactive coding and quiz widgets, using Pyodide, Lit, web workers, etc. All open source: https://github.com/pamelafox/proficient-python
I'm DongYoon, founder of Markhub (https://markhub.ink).
We're building a chat app that automatically creates and manages your to-do list right from your conversations.
I started this for a simple reason: I was tired of the soul-crushing 'copy-paste' work of moving decisions from Slack over to Notion or Jira. So much context gets lost in that process, and it just creates more "work about work."
Our core idea is simple: a chat message and a to-do item shouldn't be two separate things you have to keep in sync. In Markhub, the conversation is the task. It's not a copy; the conversation itself becomes the to-do, and all the context is automatically preserved.
Our bigger vision is to do for collaboration what GitHub did for Git. We’re not reinventing chat or kanban boards; we’re building the seamless 'workflow layer' on top that finally makes them work together.
We're currently in a private beta and would love to hear from HN users who feel this same pain. We’ve been fortunate to get early traction with large enterprise clients (including a ~$200k on-premise deal), but now we're looking for feedback from smaller, agile teams.
Any and all feedback is a gift. Thanks!
I've been working on https://stacks.camera - it's an idea about overlaying the previous picture when you're taking a photo so you can create a timelapse or animation.
For example, you can scroll through 60 pictures from my window https://stacks.camera/u/ben/89n1HJNT
Most of the challenges are around handling images & rendering, but I've also been playing with Passkey-only authentication which I'm finding really interesting.
I’m working on a help center that keeps itself up-to-date - https://happysupport.ai (landing page is German, product is English)
Creating and maintaining an up-to-date help center is a huge hassle. In many companies there is no one that really feels obligated to take care of it.
We want to optimize this process:
- Creation: Just click through your process. We take a screenshot on every click and generate a full written article with screenshots and a GIF. You can also talk while recording to add additional context.
- Maintenance: Connect to your tools (GitHub, Asana, Slack, …) and we automatically suggest changes to your docs if your product changes.
- Consumption: Users can consume the content as they like: Read the docs themselves or ask a Q&A bot.
At the moment the creation and consumption parts are already working well. Now I’m working on the maintenance part.
I'm working on inq - a real ink pen that writes on real paper while simultaneously digitizing everything you write. Specifically working on the software for our mobile and web apps.
Among other things, my team has implemented access-based sharing using web links, like Google Docs for real paper handwriting. And we've just launched Quin, our AI assistant for real paper handwriting. Super useful for getting help with math, language learning, looking up relevant facts, generating ideas, etc.
I'm one of the creators of a tool called PBRgen, which is an AI-powered online PBR material generator. The idea is that it can generate seamless materials for games and 3D animations etc. We’re still in early beta and really want to shape this with input from artists, so we are looking for artists and creators to help test and give feedback on the tool.
Here’s the link to try it out: https://pbrgen.aixpoly.com/ (limited spots available)
Let us know what works, what doesn’t, or what you wish it did. All feedback is so helpful right now.
I’m exploring two different applications of AI for education and skill-building:
1. Open-Source AI Curriculum Generator(OSS MathAcademy alternative for other subjects) Think MathAcademy meets GitHub: an AI system that generates complete computer science curricula with prerequisites, interactive lessons, quizzes, and progression paths. The twist: everything is human-reviewed and open-sourced for community auditing. Starting with an undergrad CS foundation, then branching into specializations (web dev, mobile, backend, AI, systems programming).
The goal is serving self-learners who want structured, rigorous CS education outside traditional institutions. AI handles the heavy lifting of curriculum design and personalization, while human experts ensure quality and accuracy.
2. Computational Astrology as an AI Agent Testbed For learning production-grade AI agents, I’m building a system that handles Indian astrology calculations. Despite the domain’s questionable validity, it’s surprisingly well-suited for AI: complex rule systems, computational algorithms from classical texts, and intricate overlapping interpretations - perfect for testing RAG + MCP tool architectures.
It’s purely a technical exercise to understand agent orchestration, knowledge retrieval, and multi-step reasoning in a domain with well-defined (if arcane) computational rules.
- Has anyone tackled AI generated curricula? What are the gotchas? - Interest in either as open-source projects?
Side project to enable speech-to-text functionality on my Ubuntu laptop so I can prompt AI coding tools fast:
Just vibe build a Table Viewer based on Markdown table syntax you can share via links. https://v0-markdown-to-table-view-snipped.vercel.app/ Eventually I would like to make it embeddable into LinkedIn posts.