Featured Projects

• Python Tachyon - a educational LLM inference engine to run on consumer hardware. It generates 600 toks/​s on an RTX 4060 Ti.
• C++ / CUDA fastcv - fastcv is a C++ CUDA rewrite with Pytorch bindings of the image filters in the OpenCV library.
• C / C++ InferGPT - A high performance C/C++ inference engine for GPT based architectures that runs on CPU.
• CUDA C++ Kernels - A collection of CUDA C++ and CuTe DSL kernels written for matrix multiplication.
• ML Theory Advanced ML - A collection of advanced machine learning topics implemented from scratch.