Omar Walweel
Backend & AI Engineer | Rust, Python, Distributed Systems
New to Contra
Omar is ready for their next project!
Designed, built, and shipped WeLynk solo, a production social platform live on web, iOS, and Android. Owned everything end to end: FastAPI services, PostgreSQL/Redis, React and Flutter clients, WebSocket real-time layer, deployment, and observability. The platform combines real-time chat and presence, voice and group calls, matchmaking, 10 first-party games on an in-house SDK, virtual currency, subscriptions, and internal moderation and analytics tooling. 2.1M+ lines of code across Python, TypeScript, Dart, Rust, and Luau. Zero application-caused outages since launch.
0
34
Built a real-time multiplayer 3D engine from scratch in Rust (~294K lines): deterministic Rapier3D physics at a fixed 60Hz tick, wgpu PBR rendering to native (Metal/Vulkan/DX12) and WebGPU, and custom netcode over QUIC with delta encoding, client prediction, and rollback reconciliation for 20-player sessions. Games run in a sandboxed Luau VM, safe for untrusted code, identical logic on client and server under a strict determinism contract. One Rust core compiles to web, iOS, Android, and server. Ships with its own simulator, tick-level profiler, and telemetry analyzer.
0
39
Built the behavioral matchmaking engine at the core of a production social platform: embedding-based profile scoring, a live match queue with presence-aware availability, and a timed decision window where both users choose to keep or release the match. Engineered for the race conditions that make live matchmaking hard: simultaneous accepts, mid-queue disconnects, and double-matching, resolved with atomic Redis operations. The engine learns from behavioral signals rather than stated preferences, with anti-bot and spam detection built into the pipeline.
0
25
Built the real-time messaging layer for a production social platform: WebSocket chat with typing indicators, read receipts, reactions, edits, media, and live presence tracked authoritatively in Redis across a multi-worker cluster with no shared memory. Every message passes automated safety screening and encrypted storage on the hot path, with 100% delivery at p50 71ms / p99 326ms send-to-echo, and 400 concurrent sessions held with zero drops and zero errors on a single worker. Cut send latency ~3x by parallelizing ~25 serialized round-trips and moving link-preview generation off the hot path.
0
21