Finalmouse — Aim Gods

The challenge

A hardware company's first game: a competitive, objective-based multiplayer FPS with global reach, irregular access patterns, and no existing studio infrastructure — everything from the game to the services behind it had to be architected from zero, and shipped fast.

The decisions — and why

Decision: Amazon GameLift dedicated servers + a fully serverless AWS backend (Cognito, Lambda, API Gateway, DynamoDB). Why: A launch-spike game with irregular global access patterns punishes fixed capacity — serverless made scalability, availability, and cost one decision instead of three. Tradeoff accepted: [PLACEHOLDER — authoring session]

Decision: Fix the UE4 engine's websockets library upstream (requests truncated at 300 characters, breaking authenticated WSS) and submit the fix to Epic — rather than working around it. Why: Authenticated server communication was load-bearing; a workaround would have been a permanent liability in someone else's engine bug. Tradeoff accepted: [PLACEHOLDER — authoring session]

Decision: Establish code standards, review process, and all sprint coordination as lead — for 8 engineers across 5 time zones. Why: A distributed team ships on process or doesn't ship; the 14-month clock had no room for integration chaos. Tradeoff accepted: [PLACEHOLDER — authoring session]

The outcome — with numbers

• Shipped December 2020 — 14 months from architecture to launch
• Anti-cheat in production (Easy Anti-Cheat: Hydra, Cerberus, Scout)
• Custom chat and party systems; JWT-authorized UE4 HTTP wrapper with token refresh for AWS communication
• The websockets engine fix accepted for release in future UE4 versions