Ladder Legends — SC2 Coaching Platform

The problem

Serious StarCraft II players want feedback on their games, but extracting build orders and benchmarking them against pros by hand is tedious. Ladder Legends automates the path from 'game just finished' to 'here's what to work on.'

What I built

A three-part system I designed and built end to end: a Rust/Tauri desktop agent that watches your replay folder, classifies competitive games locally, and securely uploads them; a Python/FastAPI analyzer that parses replays into timestamped build orders and coaching metrics; and a Next.js academy that stores replays, presents the analysis, and delivers coaching video content.

Highlights

• ▹Rust/Tauri desktop agent: cross-platform replay-folder detection, local s2protocol parsing, OS-keychain token storage, device-code login, background tray operation, and self-updating signed binaries.
• ▹Python analyzer reaching ~99% build-order parity with sc2replaystats — APM/EPM split, production timelines, build fingerprinting, and pro benchmark comparison.
• ▹Next.js academy on Vercel: Discord OAuth, a replay library, Mux-hosted coaching videos, a markdown-driven CMS, and product analytics.
• ▹A resilient upload pipeline: hash-based dedup, single-flight scanning, 401-triggered re-auth, atomic config writes, and Windows concurrency fixes.