JobSearchBot

A bot that scans all active listings on JobSearch.az and notifies the user on Telegram about jobs matching their profile. Vision and roadmap: VISION.md · Technical architecture: ARCHITECTURE.md

How it works

• GitHub Actions runs it 5 times a day (Baku 09/12/15/18/21) — no local machine is required.
• The scraper paginates the site's JSON API and pulls all ~2700 listings (with retry/backoff).
• There are two flows:
  • Legacy (single user): rule-based classification (jobbot/classifier.py) for the TELEGRAM_CHAT_ID in .env/Secrets.
  • Phase 1 (multi-user, Supabase): users send /start to the bot and upload a CV (PDF); each user receives matches based on their own profile. The listing corpus is shared across all users and is analyzed once, then cached.

Setup (development)

py -m venv .venv
.\.venv\Scripts\python.exe -m pip install -r requirements.txt
Copy-Item .env.example .env   # fill in the values
.\.venv\Scripts\python.exe -m jobbot --dry-run

Phase 1 setup (multi-user)

1. Create a Supabase project → run the contents of supabase/migrations/001_init.sql in the SQL Editor.
2. Add to .env (local) and GitHub Actions Secrets:
   • SUPABASE_URL, SUPABASE_SECRET_KEY
   • GROQ_API_KEY (optional — without it, the CV profile is extracted heuristically)
3. When the bot runs, it processes incoming /start + CV uploads on every cron cycle.

Privacy rule: CVs are personal data — they are only sent to providers that do not train on the data (Groq/Cohere). Details: ARCHITECTURE.md.

GitHub Secrets (Actions)

Secret	Required	Description
TELEGRAM_BOT_TOKEN	✅	BotFather token
TELEGRAM_CHAT_ID	✅	Legacy single-user chat id
SUPABASE_URL	For Phase 1	Project URL
SUPABASE_SECRET_KEY	For Phase 1	service_role secret key
GROQ_API_KEY	optional	CV profile extraction (LLM)
OPENAI_API_KEY	optional	Legacy AI classification

License

Licensed under the Apache License, Version 2.0. See LICENSE.