Competitive mini-game platform with cash prizes
A skill mini-game platform where players compete in matches with cash prizes paid through PIX: seven 2D web games in TypeScript, a web portal, an admin panel and a serverless back-end with NoSQL and relational databases, anti-cheat, an auditable ledger, ranking and infrastructure as code. In assisted operation, with a gradual launch in 2026.
The challenge
Build from scratch, in months, a platform where real money comes in and goes out through PIX and a player's score is worth a prize: cheating had to become expensive and detectable, the books had to balance to the cent, the games had to be proven games of skill rather than chance, and everything had to be operated by a small team.
The objective
Launch seven mini-games with three-player rooms, ranking, missions, a store and PIX payments, with a complete audit trail and a gradual launch without interrupting operations.
The solution
Game client on a 2D web engine with TypeScript, WebP atlases and per-game asset packs; web portal (PWA, several languages) and admin panel; serverless Node back-end with a real-time NoSQL database and a relational database (179 endpoints), a dual-write ledger with audit tables, idempotent processing and a circuit breaker on the PIX gateway; anti-cheat with HMAC-signed scores, server-generated seeds and layouts, a honeypot, replay auditing and an obfuscated build; bots filling incomplete rooms and statistical skill-versus-luck simulation; managed authentication with MFA, anti-bot verification, app attestation and identity verification; infrastructure as code (network, edge WAF, permissions, secrets), managed queues and a data warehouse; CI against emulators; a generative-AI coach and automatic rhythm-map generation from audio.
The impact
- Seven mini-games in production on a single platform, with three-player rooms, tiered MMR ranking, missions, streaks, a store and subscriptions
- 179 back-end endpoints, 253 test files and 67 recorded architecture decisions in about six months
- PIX cash flow with an auditable ledger and idempotent processing, identity verification before withdrawals and MFA for administrators
- Gradual launch with separate production and staging environments and the whole infrastructure as code

Jean C Becker
Senior Solutions Architect | AI & Machine Learning Specialist
Designed and built this work. About JBKR →
