Challenge
Aggregate job postings from 130+ wildly inconsistent sources — cached aggregators, enterprise ATS platforms, and fragmented custom career pages — and surface only the handful of jobs actually worth applying to.
AI · Full Stack
JobsAI — Autonomous Job Discovery & Application Engine
JobsAI is a fully autonomous job discovery and application pipeline. A multi-layer scraping engine ingests postings from over 130 sources using three strategies: structured portals are parsed directly from XML sitemaps and JSON-LD metadata to bypass UI changes and caching; companies on enterprise ATS platforms like Workable and Greenhouse are read straight from their backend data APIs, skipping the frontend entirely; and fragmented custom career pages are handled by extracting the raw DOM and letting an LLM normalize the chaotic markup into clean, structured JSON. Every incoming job is then parsed by a local text engine that extracts implicit tech stacks buried in prose and computes a 0–100 match score against preferred titles, experience, location, and exact technology requirements — automatically sorting jobs into matched, below-threshold, or irrelevant so the dashboard only shows what matters. A smart document router holds multiple resume variants and algorithmically picks the best one per job while generating targeted cover letters from profile metadata. Everything is tracked in a Next.js 15 application backed by a local Prisma database, with real-time pipeline metrics, per-company registries, and full scrape/automation logs.
Overview1 / 7
Solution
Engineered a three-tier scraping engine (sitemap/JSON-LD parsing, direct ATS API access, and LLM-powered DOM normalization) feeding a local matching engine that scores every job 0–100 against profile keywords, titles, experience, and location, plus a document router that selects the best resume variant and generates tailored cover letters per application.
Outcome
A personalized recruiter that runs itself: thousands of scraped postings distilled to a ranked shortlist, irrelevant jobs hidden automatically, the right resume attached to every application, and the entire pipeline — from scrape logs to daily application limits — visible in one real-time dashboard.