I built an AI that finds and scores my next job

Description and use

A pipeline that automatically collects, filters, and scores job postings, reducing noise to 12 relevant opportunities per week, improving job‑search efficiency for a career changer.

Problem

Career changers lack tailored job discovery tools that filter for specific constraints and validate postings before application.

Outcome

The pipeline processes ~1,500 postings yearly, parking 1,539 and presenting 12 high‑confidence, vetted job opportunities weekly, enabling focused application efforts.

Workflow

• Collect job postings from 16 APIs, RSS feeds, and Playwright scrapes
• Deduplicate results in a SQLite store
• Apply deterministic hard filters (title, location, recruiter spam, skill experience)
• Score survivors with an LLM against personal constraints
• Verify posting bodies to eliminate false positives
• Manually review and submit applications

Evidence

Every week my tooling sweeps around twenty job sources and pulls in a few hundred postings.

Source

Stage 2, hard elimination. Before any model sees anything, cheap deterministic filters kill the obviously wrong postings.

Source

Stage 3, LLM scoring. The survivors get read by a language model that scores fit against my honest profile.

Source

Stage 4, verification. Nothing enters the apply queue until the real posting body has been fetched and read.

Source

I kept waiting for a project impressive enough to prove I could engineer software. At some point I noticed the most honest proof was already running: tested Python, real scrapers, a dedupe store, CI on the ETL framework, and a system in daily use solving a real problem with real stakes.

Source