JLA - Job Listing app

Two projects, one system:

• job-search-automation — fetches, filters and scores remote job listings with a local LLM, persists them to Postgres, and exposes them (plus its own config) over an HTTP API.
• job-listings-app — a Nuxt UI that reads those listings and lets you browse/filter them, and edit the backend's config.yaml from a /settings page.

jla/
├── docker-compose.yml
├── job-search-automation/   # backend: scraping, scoring, Postgres, API
└── job-listings-app/        # frontend: Nuxt UI

Run everything

cd job-search-automation && cp config.yaml.example config.yaml && cd ..
cd job-listings-app && cp .env.example .env && cd ..

docker compose up -d --build

• Frontend: http://localhost:3000/jobs
• Settings panel: http://localhost:3000/settings
• Backend API: http://localhost:8000 (docs at /docs)

Populate the database with a scan:

docker compose run --rm backend python main.py

Ollama isn't part of this compose file — point OLLAMA_URL (in docker-compose.yml, or ollama.url in config.yaml/the settings panel) at wherever it actually runs: your host, or another container on your shared Docker network.

How the pieces talk to each other

job-listings-app (Nuxt)
  server/api/jobs.ts       --> GET  backend:8000/jobs
  server/api/settings.get.ts --> GET  backend:8000/config
  server/api/settings.put.ts --> PUT  backend:8000/config

job-search-automation (FastAPI)
  GET  /jobs     --> reads Postgres
  GET  /config   --> reads config.yaml
  PUT  /config   --> validates + writes config.yaml
  POST /run      --> runs the scan pipeline in the background

The browser never talks to the backend directly — Nuxt's server routes proxy it, so NUXT_BACKEND_URL is the only place that knows the backend's address.

Scheduling scans

POST /run triggers a scan on demand, but nothing calls it periodically. Wire that up however fits your setup — a host cron job running docker compose run --rm backend python main.py, a systemd timer, or a call to POST http://localhost:8000/run from whatever scheduler already runs on your shared Docker network.

See each project's own README for details specific to that half of the system.