04 — CONTRIBUTE & SHARE

Open Source

I actively contribute to high-performance inference libraries, design self-hosted developer toolings, and tinker with operating system kernels.

FAST GENERATION

LLMs & Inference Optimization

buddywhitman/sglang

SGLang is a high-performance serving framework for large language models and multimodal models.

Python

A high-throughput and memory-efficient inference and serving engine for LLMs

Python

JAX backend for SGL serving stack.

Python

AI Tensor Engine for ROCm AMD GPUs.

Python

PRODUCTIVITY ENGINE

Developer Tools & Self-Hosting

Distributed Multilingual Document Processing System Built for XX-TB Scale with persistent error logging and resume capabilities.

Python

An automatic task and event scheduler utilizing LLMs and message parsing.

Python

DOWN TO METAL

Embedded & VLSI

A RISC-V RV32I core implemented using Verilog & SystemVerilog.

Verilog