Building at the intersection of real-time systems, generative AI, and interactive environments.
I specialize in low-level system programming, cross-platform audio capture, shader development for VRChat, and AI-driven pipelines. My engineering philosophy: performance, clarity, and reliability.
Languages
Frameworks & Tools
Cloud & Infra
AI / Audio / XR
| Project | Description |
|---|---|
| 0xBitNet | Run BitNet b1.58 ternary LLMs with WebGPU — in browsers and native apps. |
| ProcTap | Per-process audio capture via WASAPI process loopback (Windows; Linux/macOS planned). Low-latency PID taps for Python. |
| OBS-WebRTC-Link | Versatile WebRTC plugin for OBS Studio — universal low-latency WebRTC input & output. |
| ai_research_dashboard | Latest AI/ML papers, models, repos, trends & news — auto-updated every hour. |
| NxPU | WGSL → NPU transpiler. Compiles WebGPU shading language to CoreML, ONNX, TFLite, and StableHLO. |
| PyETWkit | High-performance ETW (Event Tracing for Windows) consumer library for Python with a Rust core. |
| PyGPUkit | Minimal GPU runtime for Python — high-performance CUDA kernels, memory management, and LLM inference without heavy dependencies. |
| KTApple | KDE-style KWin tiling for macOS — visual tile editor, Shift+Drop placement, gap-drag resize. No SIP required. |