I ended up learning about parametric equations again, and make this visualization to document what I learned:
https://visualrambling.space/moving-objects-in-3d/
feel free to visit and let me know what you think!
The problem was that context kept disappearing between tasks. With multiple Claude agents running in parallel, I’d lose track of specs, dependencies, and history. External PM tools didn’t help because syncing them with repos always created friction.
The solution was to treat GitHub Issues as the database. The "system" is ~50 bash scripts and markdown configs that:
- Brainstorm with you to create a markdown PRD, spins up an epic, and decomposes it into tasks and syncs them with GitHub issues - Track progress across parallel streams - Keep everything traceable back to the original spec - Run fast from the CLI (commands finish in seconds)
We’ve been using it internally for a few months and it’s cut our shipping time roughly in half. Repo: https://github.com/automazeio/ccpm
It’s still early and rough around the edges, but has worked well for us. I’d love feedback from others experimenting with GitHub-centric project management or AI-driven workflows.
I’d be glad to hear what you think. If you’ve ever had to wrestle with generating PDFs or images from HTML, I hope this feels like a smoother option. Feedback, ideas, or even just impressions are all very welcome, and I’d love to learn how PlutoPrint could be more useful for you.
We take high level model code, like you'd have in PyTorch, and generate very fast GPU code. We do that without using LLMs or AI - rather, we pose it as a search problem. Our compiler builds a search space, generates millions of possible kernels, and then searches through it to minimize runtime.
You can try out a demo in `demos/matmul` on mac to see how Luminal takes a naive operation, represented in our IR of 12 simple operations, and compiles it to an optimized, tensor-core enabled Metal kernel. Here’s a video showing how: https://youtu.be/P2oNR8zxSAA
Our approach differs significantly from traditional ML libraries in that we ahead-of-time compile everything, generate a large search space of logically-equivalent kernels, and search through it to find the fastest kernels. This allows us to leverage the Bitter Lesson to discover complex optimizations like Flash Attention entirely automatically without needing manual heuristics. The best rule is no rule, the best heuristic is no heuristic, just search everything.
We’re working on bringing CUDA support up to parity with Metal, adding more flexibility to the search space, adding full-model examples (like Llama), and adding very exotic hardware backends.
We aim to radically simplify the ML ecosystem while improving performance and hardware utilization. Please check out our repo: https://github.com/luminal-ai/luminal and I’d love to hear your thoughts!
But with two additional twists:
1. It loads up historical maps from different years (right now 1 BC, 700 AD, 1000 AD, 1300 AD, 1800 AD, 1900 AD) so you can see what you would hit if you had a time machine AND you went in the direction your phone is pointing
2. Tap a country/territory for an (AI-generated) blurb on what you are pointing at
How it works: Starting from your phone’s bearing, we trace the great-circle in 200 km steps, prefilter candidate countries with bounding boxes (~5–10 instead of ~200), then check ~20 km points along each segment to catch coastlines and stop when the path first enters another country.
Great-circles (https://www.movable-type.co.uk/scripts/latlong.html) are why you can hit Australia from NYC, even though when you look at a flat map that can be hard to see.
There might be some weird stuff in the explanations, I haven’t read all 1,400 of them. If you see something weird let me know and I will update it!
The app is free and doesn’t have ads or tracking — your location and bearing are only used locally to figure out where you are and what you’re pointing at
Probably will work best if you hold your phone pretty flat :)
Thank you to André Ourednik and all the contributors to the Historical Basemaps project: https://github.com/aourednik/historical-basemaps)
That's why we built Relay: a free, browser-based tool that streamlines the ACME workflow, especially for tricky setups like homelabs. Relay acts as a secure intermediary between your ACME client and public certificate authorities like Let's Encrypt.
Some ways Relay provides a better experience:
- really fast, streamlined certificates in minutes, with any ACME client
- one-time upfront DNS delegation without inbound traffic or DNS credentials sprinkled everywhere
- clear insights into the whole ACME process and renewal reminders
Or read our blog post: https://anchor.dev/blog/lets-get-your-homelab-https-certifie...
Please give it a try (it only takes a couple minutes) and let me know what you think.
TL:DR During an online meeting, the app instantly transcribes and translates all audio you hear, and allows you to decide when you translate your voice and when you don't. It's invisible to others (like Granola), and works everywhere without any meeting bots. Try it at startpinch.com
Here's a live demo we recorded this morning, without cuts: https://youtu.be/ltM2p-SosLc
When we first launched Pinch, we shipped a video conferencing solution with a human-like AI interpreter that was an active participant in your call. Our users hold the spacebar down while speaking to the translator, and when they release the spacebar the translator speaks out to the entire room.
That design was intentional - it puts the task of context selection on the user and prevents people from interrupting each other awkwardly (only one person can press spacebar at a time). It also comes with heavy tradeoffs, namely:
* Latency - Up to 2x longer meeting lengths due to everyone hearing your full sentence and then the translation of your full sentence
* Friction with first-time users - Customers using Pinch for external communication often meet with new people each time, and we've learned of several that send out an instruction doc pre-meeting on how to join and use translation in the Pinch call. Bad signal for our UX.
* Restricting our customers to those who are meeting creators
Benefits of the desktop app:
1. It creates a virtual microphone that you can use in any meeting app
2. Instant transcription+translation means you can understand what's going on in real-time and interrupt where necessary
3. Simultaneous translation - after you start speaking, the others will hear your translated audio as fast as we can generate it, without interrupting your flow.
Over the last months our focus has been on developing a model and UX to support high translation accuracy while automating context selection - knowing exactly when it has enough words to start the translated sentence. We’ve rolled this out to the desktop app first.
We're incredibly excited to go public beta today, you can give it a try at www.startpinch.com
Cheers, - Christian
The site is free. Elecrow charges about $1 per pcb in quantities of 50 and $0.80 in quantities of 100.
No likes or followers (for now) — just gentle presence.
Built entirely by a solo engineer, including the custom operating system it runs on: OS Yamato.
What is OS Yamato?
OS Yamato is a poetic, ephemerality-first operating system designed to let digital memories bloom and fade, like flowers in the wind.
Features include:
Diaries that bloom like seasonal flowers — and wither if forgotten
Photos and videos that quietly disappear unless viewed again
Chat with ambient seasonal effects and no pressure
A soft, nature-inspired calendar with reusable schedule templates
Wind Messages — time-delayed letters that arrive months later
A 3D globe to see fleeting connections blossom around the world
Each app within OS Yamato reflects a different way of embracing impermanence in the digital world.
Learn more about the concept and philosophy: [YouTube – Write a diary, and a flower will bloom](https://youtu.be/JrqwU_N5WBA?si=kZQAadNdBGipk6gz)
Try Hanaco Weather now: https://hanaco875.com
#ShowHN #HanacoWeather #OSYamato #solodev #minimalism #digitalpoetry #webapp #ephemeral
Most metrics engines today use proprietary formats to store data and don’t use disaggregated storage and compute. Okapi changes that by leveraging open data formats and integrating with existing data lakes. This makes it possible to use standard OLAP tools like Snowflake, Databricks, DuckDB or even Jupyter / Polars to run analysis workflows (such as anomaly detection) while avoiding vendor lock-in in two ways - you can bring your own workflows and have a swappable compute engine. Disaggregation also reduces Ops burden of maintaining your own storage and the compute engine can be scaled up and down on demand.
Not all data can reside in a data-lake/object store though - this doesn’t work for recent data. To ease realtime queries Okapi first writes all metrics data to an in memory store and reads on recent data are served from this store. Metrics are rolled up as they arrive which helps ease memory pressure. Metrics are held in-memory for a configurable retention period after which it gets shipped out to object storage/datalake (currently only Parquet export is supported). This allows fast reads on recent data while offloading query-processing for older data. On benchmarks queries on in-memory data finish in under a millisecond while having write throughput of ~280k samples per second. On a real deployment, there’d be network delays so YMMV.
Okapi it is still early — feedback, critiques, and contributions welcome. Cheers !
To add to this, finding properties that meet the budgets, commuting preferences and work locations of everyone in a group is a Sisyphean task - it often ends in failure, with somebody exceeding their original budget or somebody dropping out.
To solve this I built a tool (https://closemove.com/) that:
- lets you enter between 1-6 people’s workplaces, budgets, and maximum commute times
- filters public rental listings and only shows the ones that satisfy everyone’s constraints
- shows results in either a list or map view
No sign-up/validation required at present. Currently UK only, but please let me know if you'd want me to expand this to your city/country.
This currently works best in London (with walking, cycling, driving and public transport links connected), and works decently in the rest of the UK (walking, cycling, driving only).
This started as a side project and it still needs improvement. I’d appreciate any feedback!
So we built an agentic framework that taps, swipes, types… and somehow it’s outperforming giant labs like Google DeepMind and Microsoft Research on the AndroidWorld benchmark.
We were thrilled about our results until a massive lab (Zhipu AI) released its results last week to take the top spot.
They’re slightly ahead, but they have an army of 50+ phds and I don't see how a team like us can compete with them, that does not seem realistic... except that they're closed source.
And we decided to open-source everything. That way, even as a small team, we can make our work count.
We’re currently building our own custom mobile RL gyms, training environments made to push this agent further and get closer to 100% on the benchmark.
What do you think can make a small team like us compete against such giants?
Repo’s here if you want to check it out or contribute: https://github.com/minitap-ai/mobile-use
Our discord: https://discord.gg/6nSqmQ9pQs
<Examples Queries> "Check PostgreSQL server status" "Check PostgreSQL server version and connection status" "Verify if extensions are installed" "Show current active connection count" "Show the shared_buffers configuration" "Show PostgreSQL configuration parameter for shared_buffers" "Find all memory-related configuration settings" "Show logging configuration parameters" "Display connection-related settings" "Find all timeout configurations" "Show all PostgreSQL configuration parameters" "Show top 10 slowest queries" "Show top 20 slowest queries" "Analyze slow queries in specific database" "Find unused indexes" "Analyze recent query activity" "Check index efficiency in specific database" "Check database sizes" "Find largest tables" "Show tables that need VACUUM" "Check table sizes in specific database schema" "List tables in specific database" "Check maintenance status in specific database"
I was tired of bloated invoicing apps, paywalls, and forced signups, so I built PineBill. It runs entirely in your browser: fill in products and customer details on the left, see a live preview on the right, then export/download a PDF. No ads, no tracking, no account.
I first shared it on Reddit and got a bunch of thoughtful feedback; it’s now at ~3k monthly active users. I’d love your critiques, bug reports, and feature ideas.
Reddit post: https://www.reddit.com/r/SideProject/comments/1ksnikx/i_made...
GPT-5 by itself gets 65.0%, Sonnet 4 64.8%, but randomly switching at every step gets us 67.2%
This result came pretty surprising to us. There's a few more experiments in the blog post.
Simple Kanban boards, easy login that takes you straight to it.
It uses socket.io for data xfer after login which was fun to setup and also makes it work well.
Totally free, any feedback would be cool or if you wanted some info on the tech just ask.
Feedback welcome on UI and categorization quality.
I noticed that many vibe coders have a problem with building their apps. It's nice and easy when you're starting, but when you need to launch a production-ready app... The fun stops.
That's why I built DeVibe: to connect vibe coders with developers.
- Vibe coders post jobs; devs apply, and get accepted, or - Devs create services, which vibe coders can book immediately.
Goals: make it a go-to place for vibe coders needing help, create more dev jobs in a tough market, and help end users get safer/stable apps.
Would love to hear your feedback!
Remember the dating section in newsletter ads?
No algorithms. No swipes. No signup.
Everyone is trying to put even more "AI" into their products. I tried to do the opposite with https://1999date.com .
Find interesting profiles in your city or submit your own profile to get found.
Would be really happy for suggestions and feedback! What would you add or improve?
See you,
David
So I built Socks2Stocks, a simple tool that automatically estimates fair value using a discounted cash flow (DCF) calculator.
Originally, I overbuilt it with too many half-baked features. Everything felt wrong, so I scrapped the whole thing and rebuilt it from scratch with just one feature: the DCF calculator I always wanted when I first got into investing.
How it works The app pulls financial data through an API, runs free cash flow projections, and automatically outputs a fair value estimate. The cash flow projections are based on the average growth rate of FCF over the past 10 years and then projected forward. Users can, of course, enter their own assumptions.
I’d love any feedback, and to hear whether you think people would actually use this. Thank you for reading.
I was getting tired of all the news websites being incendiary and emotionally manipulative so I built this using AI. The goal is to keep the reader informed and nothing else.
Feedback welcome!
See something you want → Log it → Choose resist or buy → Watch your savings grow
No budgets. No lectures. Just awareness that builds discipline.
Please give your feedback to improve it further
I’m Lily, a solo indie maker, and I’ve been working on Textideo — an AI video generator that turns text prompts into short videos.
Why I built this
I’ve always found creating videos challenging. Traditional editing tools are heavy, expensive, and require skill. On the other hand, most AI tools I tried were either locked behind subscriptions or required publishing outputs publicly. I wanted something simple: pay-as-you-go, private, and accessible to anyone.
How it works
You enter a short text prompt (e.g., “A cat walking on the grass at sunset”).
The system runs it through multiple text-to-video models (Google Veo 3, WAN, etc.).
Within 30–60 seconds, you get a short MP4 clip you can download.
No subscription, no login required for trying the demo.
Under the hood:
Backend is built with Python (FastAPI) + PyTorch + FFmpeg for video rendering.
Frontend is in React + Tailwind, optimized for minimal UI friction.
Videos are generated via a credits system instead of recurring payments.
What’s different
No subscription lock-in — buy credits only when you need.
Privacy first — you don’t need to make your outputs public.
Multiple models in one place — test different AI video generators without juggling accounts.
Current status
The project is in beta. You can try generating a short video right on the homepage. I’d love to hear your feedback — technical, UX, or ideas for improvement.
Thanks for reading, and I’m happy to answer any questions!
Magic Translate is designed to work with images with burned-in text. All you need to do upload is your original image, and Magic Translate will do the rest. No design files required.
This is the first launch, right now it only supports use on Desktop, but I'm looking to add a mobile experience soon.
The primary use case I had in mind was for marketers who wish to localize their content, but don't necessarily have the design team nor the time to hand-translate their creative assets. This app is singularly focused on image translation, rather than an entire suite of design tools.
It's free to try out, feedback and suggestions are much appreciated!
Check it out:
https://smart-search.kodex-ai.com/
Thoughts or feedback?
What you can do: Train miners to gather gold and stone Convert stone into gold for more resources Recruit units like shielders, archers, barbarians, and wizards Build and upgrade defensive towers Lead your army to destroy enemy bases and free your people from the Red Empire Features: 3 difficulty levels: Easy, Normal, Impossible Works on desktop, tablet, and mobile browsers Also available on Google Play and App Store No downloads required, play instantly online Family‑friendly, suitable for all ages Why I made this I grew up playing simple stickman games and wanted to create a modern version with strategy depth but the same instant accessibility. The game is built to run smoothly in the browser and be playable even on school or work networks (unblocked).
Would love feedback from the HN community on gameplay balance, performance across devices, and ideas for future unit types or mechanics.
You can try it here: https://warofsticks.app/
What it is:
Players submit interesting claims such as "I think bitcoin can reach $1MM." Other players vote on whether each submission is Truth or Myth.
Why this works:
The best part is the community creates all the content. I've been amazed by the creative myths people invent and the bizarre-but-true facts they dig up. Some submissions have stumped 90%+ of players.
What surprised me:
Players are incredibly good at crafting believable myths. The voting patterns reveal fascinating insights about what kinds of "facts" people instinctively trust vs. doubt. The HN crowd tends to be skeptical of everything - curious to see how you perform compared to other communities I've tested this with.
Play here: https://truth-wave.lovable.app/ Would love feedback on the mechanics and any interesting submissions you come up with!
Will incorporate more features as statements increase.
I built Web Server Components, a minimalist (~1KB min+gz) TypeScript library that enables SSR + hydration with Custom Elements using just a simple decorator and base class — no frameworks, zero dependencies.
At its core:
withWSC decorator auto-registers your element.
WSC is an isomorphic HTMLElement wrapper — so you can safely call .render(...) on the server and append the same element client-side.
Highlights:
~1KB min+gz
Zero dependencies
Standards-only: Custom Elements + Declarative Shadow DOM
SSR + minimal hydration with no framework lock-in
Live demo: https://1kb.alexandre-giordanelli.workers.dev/
Repo: github.com/alexandregiordanelli/web-server-components
The goal is to show that you don’t need a framework to get SSR + hydration with Web Components — just a decorator and a lightweight HTMLElement wrapper.
Would love feedback, comparisons with other minimal SSR approaches (htmx, Lit, etc.), and ideas for improvements. Thanks!
So I built a React starter kit! It gives you the building blocks to kick start your trading application with customisation and core trading components.
Upcoming Integration: MT4/MT5/Forex by 2026.
To confirm my experience, I looked through 402 open source healthcare AI repos on GitHub and found that almost half the infrastructure tools are just trying to solve interoperability problems.
So I built HealthChain to make working with healthcare data less painful. It lets you connect AI pipelines to data sources like FHIR APIs with built-in data converters and validation in a few lines of Python.
For more context and some pretty graphs on the Github survey results: https://jenniferjiangkells.substack.com/p/healthchain-buildi...
Happy to answer any questions and would love to hear from anyone working in healthcare AI or just curious about the space!
I’ve been building StrangerMeet , a lightweight text and video chat platform inspired by Omegle but rebuilt with a modern stack. Omegle shut down last year, leaving a lot of people looking for real-time anonymous chat options. StrangerMeet tries to fill that gap with a cleaner design and some improvements:
Video + Text Chat – One-click random connections
Continue Chat – Option to reconnect with someone if both users agree
Group Rooms – Public or private rooms for shared conversations
Temporary Media & Voice – Share images, voice notes, or short videos that auto-expire
Mobile-Friendly UI – Fullscreen, app-like experience on phones
The backend is Node.js with WebSockets + Redis for matchmaking, and the frontend is React + Tailwind. I’ve tried to keep it fast, simple, and privacy-conscious.
I’d love your feedback on:
Scalability of the architecture
Ideas for keeping moderation effective but lightweight
Features you think an Omegle-style platform should (or shouldn’t) have in 2025
It’s live here: https://stranger-meet.com
Thanks!