JobFinder scrapes companies' job boards, filters out the noise with an LLM, and hands you a scored shortlist with pros and cons, so you can spend your time doing literally anything else.
An agentic pipeline that turns hundreds of postings across every company you care about into a short, scored list.
Scrapers pull every open role straight from each company's job board API; no headless-browser flakiness, no manually paging through careers sites.
A filtering agent skims the full list of titles and locations and drops anything that's clearly not a fit, so the expensive analysis only runs on real candidates.
An analysis agent reads each remaining posting against your resume and preferences, scoring the fit and writing out the trade-offs.
Jobs land in a searchable, filterable list with a match score and pros/cons on every card. You only read the ones worth reading.
Analysis runs against whatever model you point it at: fully offline with Ollama, or hosted with an Anthropic or OpenAI API key.
Run everything offline on your own hardware, with zero API costs. gpt-oss:20b on a single RTX 4090 comfortably chews through a full overnight run of every company.
Anthropic's models. Paste in an API key during onboarding and pick any model your key has access to. API keys never leave your machine unencrypted.
platform.claude.com
OpenAI's models, and the origin of the gpt-oss family that works so well offline. Works the same way: one key, any model.
Clone it, start it, and walk through onboarding in your browser: pick your companies, your model, and paste in your resume. Then kick off a run and come back to a scored list.
# grab the code git clone https://github.com/chrisdothtml/ai-job-finder.git cd ai-job-finder # install deps (Volta auto-picks the right node + yarn) yarn install # build the UI & start the server yarn app:build && yarn app:start
