Abdurrahman Muhammed

I’m a C++ systems engineer with industry experience building production systems for EA’s F1 series. I focus on performance optimisation, ECS architecture, multithreading, and low-level systems programming—especially where cache, memory layout, and real-time constraints matter.

Alongside my engineering work, I contribute as an AI Evaluation Contributor at Outlier, analysing C++ implementations generated by LLMs. I identify correctness issues, undefined behaviour, and inefficiencies in memory and performance, then justify preferred approaches using structured rubrics across edge cases and low-level STL behaviour.

At Electronic Arts (Codemasters), I designed modular gameplay systems in C++17 across a 5M+ line codebase using archetype-based ECS with component pooling. I improved iteration speed by 30% with live-reload tooling, increased engine stability by 15% using Superluminal/PIX profiling and memory diagnostics, reduced per-frame allocations by 22%, and eliminated 8 critical leaks via custom arena allocators; I also achieved a 35% cache hit improvement through AoS-to-SoA conversion and hot/cold data splitting.

I also strengthened reliability and performance by debugging 40+ high-priority issues, including race conditions in multithreaded telemetry, state corruption in networked gameplay, and cache thrashing across hot gameplay, rendering, and AI paths. My project work reflects the same priorities—e.g., a 230x speedup GPU N-body simulation with shared-memory tiling, and a multithreaded match engine delivering 40,000 orders/sec and 12µs 95th-percentile latency using lock-free SPSC queues—plus an accessibility-focused game engine built with archetype ECS running 500+ entities at 60 FPS.