Electron + FastAPI desktop overlay with a 3-layer rune recommendation engine (player history → defaults → high-elo data)
A League of Legends desktop companion that detects champion select via the LCU API and automatically recommends and imports the best rune page with one click. The recommendation engine checks the player's own win history first, falls back to curated champion/role defaults, and finally to aggregated Master+ statistics. An always-on-top Electron overlay shows the recommendation instantly without disrupting gameplay.
LoL desktop overlay with a learning rune recommendation engine.
• Detection: Electron main process polls the League Client Update (LCU) API to detect champion select sessions.
• Recommendation engine (3 layers): Layer 1: player rune history with win/loss outcomes; Layer 2: curated default pages per champion/role; Layer 3: high-elo (Master+) aggregated rune statistics.
• One-click import: recommended rune page is imported directly into the LoL client via the LCU proxy endpoint.
• Stack: Electron + TypeScript frontend; Python + FastAPI backend; PostgreSQL via SQLAlchemy; Jest (frontend) + pytest (backend).
• Infra: Docker Compose (Postgres + FastAPI); Alembic migrations; .env-based config.