stfurkan

Hi again :) I updated the engine and 27B model should work in the demo[1]. We can still consider it as experimental but I am trying to make it better and efficient.

1. https://stfurkan.github.io/bitgpu/examples/chat.html

Thank you :) I wanted to create it from zero myself so I can control it better. e.g. I started with ONNX support but now it support gguf models too. It might be a good idea to add some popular connectors in the future. Good luck with your project :)

Hello, I sent you a LinkedIn connection request. We can chat first and schedule a meeting. Thanks

Hi, the engine itself already has tool calling capabilities but if you're asking for the aidekin I didn't add tool calling there on purpose.

It needs webgpu support but I haven't had a chance to test with different devices.

Hi :) There are some smaller embedding and re-ranking models (I am currently using some of them on my aidekin project) but having it like bonsai models will be great to have.

Thank you for trying it :) I just added the 27B support 1 hour ago. Let me check and try to fix it. All other models should work as expected but if you have issues with them too, I can take a look.

I'm currently working on an open-source engine [1] exactly for this purpose. If anyone wants to try or has any suggestions, I'm happy to listen :)

1. https://github.com/stfurkan/bitgpu

2. https://aidekin.com --> this is one of my projects that's currently using bitgpu engine

It looks great, thank you! I'll see if I can use it for my in browser AI assistant project's ( https://aidekin.com ) voice part. It's currently using Nemotron-3.5-ASR and supertonic-3 but overall it requires 1.2gb download.

aidekin: an open-source, client-side AI assistant you drop onto any website. Your visitors get a private voice and text assistant that runs entirely on their own device via WebGPU

https://aidekin.com

https://github.com/stfurkan/aidekin

Thank you :) Currently I am not planning to have hosted fallback Q&A but it's a nice idea. It should only download the ~300mb initially once and then use the cached model.

I see, thanks for the feedback :) It'll be a nice addition to the game.

Thank you for the feedback :) Letting players play before signing up might not be ideal for this case but I probably need to make the landing page better.

https://duckville.town

You play a duck in a small shared town. You pick a job, pay rent, post on a Twitter-style feed, vote in local elections. The simulation keeps running when you close the tab. No PvP, no loot boxes, no combat. Playtime is a few minutes a day by design.

Thank you :) There is a simple map currently but it can be improved. You can see it on this page after you sign in -> https://duckville.town/city

How can I make it better? I'd be happy to hear your feedback.

https://duckville.town

You play a duck in a small shared town. You pick a job, pay rent, post on a Twitter-style feed, vote in local elections. The simulation keeps running when you close the tab. No PvP, no loot boxes, no combat. Playtime is a few minutes a day by design.

Hello, I built it. I'm glad you like it. I'd be happy to hear your feedback

Yes, you can find the instructions here --> https://github.com/stfurkan/pi-llm

Currently I get around 2 tok/s on the 4B model, and 4-5 tok/s on the 1.7B model.

I specifically chose PrismML's 1-bit models because their tiny size allows them to actually fit on smaller hardware like the Pi. The 1.7B model is great for basic tasks and tool triggers, while the 4B model seems reasonable for some daily tasks, though it's much slower on this setup. If you try these models on your M1 Max, I assume they'll run incredibly fast. I previously tried them on a VPS and the inference speed was really good for my experiment.

Hi HN,

I spent the weekend experimenting to see if I could get a proper LLM running locally on an old Raspberry Pi 4 (4GB), and more importantly, if I could get it to interact with the physical world.

I ended up using PrismML's new Bonsai models. Because they are genuinely 1-bit (trained from scratch at 1-bit, not quantized down to 4-bit), they actually fit. The 4B parameter model is ~570 MB, and the 1.7B is ~240 MB.

I loaded them through llama.cpp's router mode. I get around 2 tok/s on the 4B model for better reasoning, and 4-5 tok/s on the 1.7B when I just need speed. I tried Gemma 4 E2B first, but it was just too slow on 4GB of RAM.

The fun part: I wired up a cheap TM1637 4-digit display to the GPIO pins. Since Bonsai supports native tool calling, I wrote a small Python proxy that injects an update_display function into requests. When the model decides to use the tool, the proxy catches the streaming call, extracts the text, and drives the display. You can tell it to "show 1453" and it physically lights up.

It’s definitely just a weekend project (7-segment displays can't render W or M, self-signed certs, etc.). The code and setup scripts are all in the repo.

I’m thinking about adding servos or sensors next. Would love to hear your thoughts or see if anyone else is building edge AI hardware projects!

I've been experimenting with AI Agents to see how far I can push them across different types of complexity (from dashboards to client-side video encoding).

Trip Replay (https://tripreplay.app) - A client-side travel map animator where I successfully got the AI to implement complex D3 projections and WebCodecs logic.

Krypto Markets (https://krypto.markets) – A crypto dashboard built purely in "Agent Mode" to test how fast I could ship a data-heavy UI.

Gez.la (https://gez.la) – My old COVID-era open source virtual tour database project that I used agents to fully refactor and modernize from a legacy stack.

Hello,

I just launched Trip Replay, my last project of 2025.

There are already tools that create animated travel maps, but they are almost all paid apps. Some charge per export, others require a subscription. I wanted something I could use myself without friction. So I built a free, web-based alternative.

How it works: It runs entirely in the browser using Next.js and the Canvas API.

* Rendering: It draws the map (using D3-geo) and the path frame-by-frame on a hidden canvas. * Encoding: Instead of a server-side render, it uses WebCodecs to encode the video directly on the client. * Privacy: Since it's client-side, no data leaves your device (except location search via API).

The output is a 1080p vertical video (9:16) ready for social media.

I launched it on Product Hunt today as well if you'd like to support it there: https://www.producthunt.com/products/trip-replay

The app is live here: https://tripreplay.app

Would love to hear your feedback!

I built a real-time cryptocurrency dashboard using AI agent mode (mostly Claude Opus 4.5) in less than 2 days.

Live demo: https://krypto.markets

Features: - Real-time prices via Binance WebSocket - TradingView-style candlestick charts (1m to 1D timeframes) - Drag-and-drop layout with save/sync - Price alerts with browser notifications - Command palette with CLI commands (⌘K to add coins, set alerts, etc.) - Fully responsive

Tech: Next.js 16 (App Router), React 19, TypeScript, Tailwind CSS 4, Zustand, Lightweight Charts, Drizzle + Turso, Better Auth

I wanted to test how far I could push AI-driven development. Claude handled WebSocket integration, responsive design, chart implementation, auth, and debugging. My role was mostly directing, reviewing, and making design decisions.

The result surprised me. What I expected to take weeks was functional in under 48 hours. Not everything was perfect on the first try, but the iteration speed was remarkable.

Curious to hear HN's thoughts on AI-driven development and the project itself.

Hi, I think spending ~$1 for fartman.fun is enough for this project :)

Hi everyone,

I wanted to experiment with Claude AI, so I created Fartman, a 2D open source game where you fight enemies with fart and spit!

Check out the code on GitHub: https://github.com/stfurkan/fartman-game

If you like it, please support it on Product Hunt: https://www.producthunt.com/posts/fartman

Hey,

I created an experimental social media which only hex code posts can be shared. I'd like to hear your thoughts :)

Also if you like the project, you can upvote it on Product Hunt :)

https://www.producthunt.com/posts/heks-social

I run https://gez.la on Vercel's free tier.

https://gez.la - Open Source Virtual Tour Database

https://gez.la - Open Source Virtual Tour Database

https://github.com/stfurkan/gez

https://gez.la - Virtual Tour Database

https://github.com/stfurkan/gez