About

I'm Alexander Lee. VoidIndex is where I document my work and learning across modern AI systems.

What I write about

• Model inference — serving architectures, scheduling, KV-cache systems, quantization, and distributed inference.
• High-performance kernels — CUDA, Triton, memory hierarchies, tiling, and performance analysis.
• Agentic reinforcement learning — training, evaluation, reasoning, tool use, and reliable autonomous agents.

My goal is to turn implementation details, experiments, and research papers into practical technical notes.

Elsewhere

You can find my projects and open-source work on GitHub.