DataSalon.ai - Discover datasets worth training on.

Hey Product Hunt! 👋 I'm the maker of DataSalon — a dataset discovery platform for ML practitioners. The problem: Finding the right training dataset takes hours. You open 6 browser tabs, read inconsistent descriptions, can't tell which is trustworthy, and still miss the ones you didn't know existed. And when you truly can't find it — there's nowhere to ask. DataSalon solves this five ways: 🌐 One place for 40+ platforms — Kaggle, Hugging Face, Zenodo, data.gov, Papers with Code and 35+ more, aggregated and deduplicated. No more tab-juggling. ⭐ Quality you can trust — Every dataset scored on 4 dimensions (Description · Source · Reputation · Access). Spam filtered, duplicates merged. You only see what's earned its spot. 🧠 AI that fills the gaps — Raw metadata is messy and uneven. Our AI pipeline normalizes every dataset into a clear title, structured summary, and unified taxonomy — so 300K+ datasets finally speak the same language. 🧭 From search to discovery — Keyword search is table stakes. What you can't search for are the combinations — "Synthetic weather data for autonomous driving perception", "Multi-lingual legal contracts with annotations". We surface those as curated Topics, so you find what you didn't know you needed. 🤝 A community for data supply & demand — Can't find it? Post a request. Have it? Post an offer. The community bridges the gap our aggregation can't cover. What's next: Quality radar charts, dataset subscriptions, and an agent interface so your AI assistants can query DataSalon directly. Shipping in beta — feedback and "this platform is missing" shouts are all very welcome 🙏 🔗 https://datasalon.ai