Resume Screener & Talent Search
The problem. One job post brings 300 to 1,000 resumes. Recruiters spend days reading them and still miss good people.
What it does. Upload a job and a batch of resumes. A pool of Go workers reads them in parallel. The AI pulls out skills and years, each resume gets a score with a reason, and you can chat with the whole pile.
- Parallel worker pool you can resize live
- Retries, circuit breaker and cancel
- RAG chat that names the candidate for every fact
- 18 MB Docker image, race-tested
No login needed
// how a resume moves through it
upload ─▶ MySQL queued
dispatcher ─▶ chan (buffered)
└─ full? wait (back-pressure)
worker × N ─▶ embed chunks
─▶ LLM → JSON schema
─▶ score in Go
└─ retry + back-off, circuit breaker
chat ─▶ parallel vector search
─▶ streamed answer + sources