A full-stack documentation search engine that imports and cleans official web-development docs, then returns fast, relevance-ranked results through an indexing and search pipeline I built from scratch.
Built import pipelines for sitemap, llms.txt, and HTML indexes across Next.js, React, Tailwind CSS, and Prisma, extracting clean article content with Cheerio.
Designed and implemented the entire search engine myself: custom tokenization, a persisted PostgreSQL inverted index, corpus statistics, and BM25 ranking without Elasticsearch, embeddings, or database full-text search.



