GitStars

👋 I’m Dmitrii, an AI Engineer and Citizen Scientist. Like many of you, I treat GitHub Stars as bookmarks. Over the years, I collected a lot of them... and never looked back. I realized this wasn't just digital hoarding - it was a dataset of my professional evolution. So, I processed ~1TB of public GitHub data (4M+ developers) to build an engine that understands semantic similarity between repos and people. Why you should try GitStars: 🧠 The "Who are you?" Test: Enter your username and see your Skill Radar. Are you 80% AI Researcher and 20% System Engineer? The model classifies your interests instantly. 👯 Find your Dev Soulmate: The app compares your vector with famous developers (like Andrej Karpathy, George Hotz, etc.). See who matches your coding style and explore their repos that fit your stack. 💎 Find Hidden Gems: Unlike standard trending pages, this finds niche libraries that are semantically close to what you already like, filtering out what you've already seen. 🔗 Share & Compare: Generate a serverless link (encoded in the URL) to challenge your friends and colleagues: "We match 95% — we should hack together!" Under the hood: - It’s a Client-Side app. No backend. At all. - We use compressed embeddings (~70MB) and run Vector Search via WebAssembly directly in your browser. - Full sources and embeddings are available. Give it a try: 👉 Have some GitHub Stars ? Type your GitHub username. 👉 Check your "Similar to you" section. 👉 Drop a screenshot of your top match in the comments! 👉 Share a link so other hackers can compare them to you :)