Live2026
Market Fit
Ask the engineering job market and get cited answers from real postings.
A retrieval-augmented search over a live corpus of engineering job postings. Ask what companies want and get a concise answer grounded in real listings, with the exact postings it used shown as sources. A scheduled Nest job ingests postings from Greenhouse, Lever, and Hacker News, embeds them locally, and indexes them in pgvector; a LangChain.js pipeline retrieves and answers.
Highlights
- RAG on LangChain.js: text splitting, local embeddings, PGVectorStore retrieval, and an LCEL answer chain
- Self-updating corpus: a scheduled Nest job pulls Greenhouse, Lever, and HN 'Who is hiring' postings
- Cited answers: every response shows the retrieved postings and their similarity
- Free to run: on-device embeddings and pgvector on the existing Postgres, no paid vector DB
Stack
Next.jsNest.jsLangChain.jspgvectorPostgreSQLTypeScript