I build backend systems that have to hold up under real conditions — multi- tenant isolation that can't leak, stores that can't lose data on a crash. If it hasn't been tested against failure, I don't consider it done.
Right now: QuorumDB — a distributed KV store, built twice with two
replication strategies (quorum+LWW and Raft) to feel the actual tradeoff.
QuorumDB — distributed KV store, quorum+LWW and a from-scratch Raft implementation, chaos-tested under live node failures.
HiveCRM — multi-tenant SaaS CRM with tenant isolation enforced at the repository layer.
TaskWave — real-time collaborative Kanban boards over STOMP/WebSocket.
FraudShield — composable rules engine for transaction fraud detection.
GateKeeper — API gateway with JWT auth, RBAC, sliding-window rate limiting.