Find expensive code paths and reduce real frame-time cost using profiling, data-oriented changes, algorithmic improvements and low-level optimisation where justified.
C/C++ profiling and optimisation
Data layout, memory access and cache behaviour
Threading and systems-level bottlenecks
x86, x64, ARM, ARM64 and MIPS experience
GPU & Rendering Optimisation
Investigate rendering workloads and GPU bottlenecks, then target the causes that materially affect performance on the device.
Frame and GPU capture analysis
Draw-call/state overhead
Shader and rendering-path cost
Platform-specific rendering constraints
Platform Performance
Optimisation for a port or difficult target often requires balancing CPU, GPU, memory and platform-service constraints together instead of tuning one subsystem in isolation.