Automated Telegram → OCR → Pick Intelligence Pipeline

This project monitors sports-handicapper Telegram channels, extracts picks from image posts, and converts them into structured data for analysis.

• Transparency: all steps from capture → parse → store → report are inspectable.
• Safety: complies with Telegram ToS (read-only user account, throttled API calls).
• Expandability: modular layout ready for future AI models, APIs, dashboards.
• Reproducibility: deterministic Docker build for Linux/Mint/Ubuntu.

Phase 1 → Capture & Parse
Phase 2 → Scoring & Confidence Modeling
Phase 3 → Dashboards & Reporting