Designed and built a custom real-time rendering engine in C++20 and MASM x86_64, overcoming the low-level integration challenge of linking hand-written assembly (Win64 ABI, shadow space, manual UTF-16 / window structs) with modern C++20 modules into one clean pipeline.
Implemented a full DirectX 12 deferred renderer (G-buffer, shadow mapping, SSAO, volumetric fog); solved the lack of MSAA on a deferred G-buffer by implementing Temporal Anti-Aliasing (jitter + motion-vector reprojection + neighborhood clamping).
Wrote custom HLSL shaders for Screen-Space Reflections, solving ray-scene intersection with a depth-buffer raymarching/raycasting algorithm that reuses the G-buffer instead of costly ray tracing.
Eliminated a CPU draw-call bottleneck by moving to GPU-driven rendering (unified buffers + a single ExecuteIndirect) and vectorized hot math with SIMD (AVX2), improving scalability to complex scenes.