giggsodocker/reveilio repository overview

⁠reveilio

LLM-powered resume-to-job-description matching with an interactive web interface.

Upload a JD and resumes (or a zip of many), pick a provider, get ranked candidates with explainable scores, strengths/weaknesses, interview questions, and downloadable PDF reports. Optional database persistence for SQLite, PostgreSQL, MySQL, or MongoDB.

⁠Quick start

bash docker run -p 8501:8501 -v reveilio-data:/data giggsodocker/reveilio:latest

Open http://localhost:8501⁠.

⁠With docker compose

yaml
services:
  reveilio:
    image: giggsodocker/reveilio:latest
    ports:
      - "8501:8501"
    volumes:
      - reveilio-data:/data
    environment:
      - GEMINI_API_KEY=${GEMINI_API_KEY:-}
      - OPENAI_API_KEY=${OPENAI_API_KEY:-}

volumes:
  reveilio-data:

docker compose up

⁠Features

⁠Pages

1. Configure — pick LLM provider, enter credentials, optionally connect a database.
2. Analyze — upload JD + resume(s), view ranked results, optionally save to DB.
3. Database — browse / filter / delete stored analyses.
4. Reports — download PDF reports.

⁠Environment variables

All optional. UI form values override env vars.

⁠Volumes

Wipe everything: docker compose down -v (or docker volume rm reveilio-data).

⁠Ports

⁠Connecting external databases

PostgreSQL / MySQL / MongoDB not bundled — bring your own.

• macOS / Windows: use host.docker.internal to reach host DB.
• Linux: use host bridge IP (172.17.0.1) or join the same Compose network.
• Managed (RDS, Cloud SQL, Atlas): use the public hostname.

⁠Image internals

• Base: python:3.11-slim
• antiword installed for legacy .doc parsing
• reveilio library installed with [all-db] extra (every backend works without rebuild)
• Streamlit + pandas for the UI
• Healthcheck: GET /_stcore/health every 30s

⁠Security

• No authentication. Do not expose port 8501 publicly. Use a reverse proxy with auth, a VPN, or localhost only.
• API keys live in Streamlit session state — never written to disk.
• Uploaded files written to tempfile paths inside the container, cleaned by OS.

⁠Tags

⁠Library vs. image

This image bundles the Streamlit UI built on top of the reveilio Python library. If you want to integrate reveilio into your own Python code, install from PyPI instead:

bash pip install reveilio

The image and the wheel are independent distribution channels — the wheel contains only src/reveilio/ and has no Streamlit dependency.

⁠Links

• Source code: https://github.com/roosterhr/reveilio⁠
• Documentation (wiki): https://github.com/roosterhr/reveilio/wiki⁠
• PyPI package: https://pypi.org/project/reveilio/⁠
• Issues / bugs: https://github.com/roosterhr/reveilio/issues⁠

⁠License

MIT