3.3.2 AI project 2 JobRadar

← 3 Agentic AI projects

Complete project ecosystem

For details see

• Documentation: #606_ai_ides_.docx
• Repo 602-2_D1-tool-LLM

Demo 2 jobradar_unified.py (what it does so far 26.0506)

Single app that combines:
1. Ingest — Pulls jobs from Greenhouse (configured companies), normalizes and dedupes into MongoDB, optional LLM scoring (OpenAI primary, Anthropic fallback) with a simple cost tracker.
2. Gmail path (CLI) — With GMAIL_* set, the same script run ingests LinkedIn-style job alerts from a Gmail label over IMAP, parses, dedupes, scores, and writes into the same jobs collection.
3. API (uvicorn jobradar_unified:app) — FastAPI with Google OAuth (session + email allowlist) protecting job routes. Exposes health, jobs (list/detail), status updates, summary, /login / /me / /logout.
4. Digest (Step 7) — GET /digest/preview builds the digest text without sending. POST /digest/send emails it (SMTP to Gmail by default, or Resend via DIGEST_SENDER=resend — S7e).
5. Google Doc digest (Step 7f) — POST /digest/google-doc creates a new, timestamp-titled Google Doc in Drive with the same digest body, using OAuth tokens with Docs + drive.file scopes; optional DIGEST_DRIVE_FOLDER_ID and title prefix. Config is env-driven (Mongo, LLM keys, Gmail, digest, Resend, Drive folder, Google OAuth secrets, etc.).

Cursor generated.

26.0515