InferenceX
Open-source platform that continuously benchmarks SGLang, vLLM, and TensorRT-LLM across NVIDIA, AMD, and Google TPU hardware, so results stay current as the stack changes.
llm-inferencebenchmarking
PythonSGLangvLLMTensorRT-LLM
Projects
Things I've built and worked on.
Excalidraw Toolkit
Edit Excalidraw diagrams with Claude Code or Codex, review the changes in the same workspace, and export files you can keep editing.
developer-tools
JavaScriptMCPClaude CodeCodexExcalidraw
SGLang Omni
Multi-stage pipeline framework on SGLang for models that take and produce multiple modalities at once, with an OpenAI-compatible server and Real-Time API support.
llm-inference
PythonSGLangPyTorchDocker
SGLang
High-performance serving framework for large language and multimodal models.
llm-inference
PythonPyTorchCUDA
TIPs
Interactive visualization tool for minimizing language deprivation in deaf and hard-of-hearing children.
visualizationresearch
PythonReactllm
Personal Website
This site, built with Next.js, TypeScript, and Tailwind CSS. Blog, project showcase, and responsive design.
web
Next.jsTypeScriptTailwind CSSshadcn/ui