I Built a Free Patent Search Engine with 3.5M US Patents — No Login, Powered by SQLite FTS5

I'm a patent lawyer who started coding in December 2025. Today I'm launching a free patent search engine covering 3.5 million US patents (2016–2025). Try it here: patentllm.org No login. No ads. No API keys. Why I built this Existing patent search tools fall into two camps: Google Patents — simple but no filtering, no relevance tuning Commercial tools (PatSnap, Derwent) — powerful but $10K+/year I wanted something in between. Fast full-text search across titles, abstracts, and claims — with CPC classification filtering and smart relevance ranking — for free. The tech stack (intentionally boring) Component Choice Why Database SQLite (74GB, single file) Zero ops. No Postgres. No containers. Full-text search FTS5 + BM25 Built into SQLite. Handles 3.5M records without breaking a sweat. Ranking 70% BM25 + 30% recency Newer patents get a boost. Patent relevance isn't just keyword matching. Web framework FastAPI + Jinja2 Server-side rendering for SEO. Every search result has a unique URL. LLM