Ho Ko
Redefining human-computer intelligence through on-device LLMs, 3D graphics, and custom agent runtimes.
About Me
I am an AI/ML Software Engineer with cross-domain experience spanning astronomy, weather forecasting, economics, healthcare, NLP, and computer vision. I work across the entire model lifecycle — from training and distilling deep models, to architecting highly efficient inference engines, to scalable, high-throughput serving.
I bring that same intelligence directly on-device through WebGPU, Three.js, and local LLMs — building sandboxed systems and custom peer-to-peer protocols that deliver responsive, privacy-preserving, and completely offline experiences.
Skills & Expertise
Project Explorer
Left click + Drag to rotate. Scroll to zoom. Hover over nodes to see titles. Click a star to inspect the project details.
Chat with Virtual Me
Have a conversation with an AI assistant loaded with my professional timeline, research interests, and project knowledge. To preserve maximum privacy and keep everything open, this conversation runs locally in your browser via WebGPU.
Select In-Browser LLM Model
Runs entirely on your device's GPU via WebGPU. Select a model to download and compile.
LFM 1.2B
Quick Thinking model
Size: ~900MBGemma 4 E2B
Smarter assistant (MediaPipe LiteRT)
Size: ~2GBHi! I am the virtual avatar of Ho Ko. Ask me anything about my experiences, projects, or background, and I will search my local index to answer you!
Experience & Education
Education
M.S. in Computer Science
Specialization in AI & ML (Completed post-transfer)
GPA: 4.0/4.0M.S. in Electrical & Computer Engineering
Focus on Machine Learning & AI (Transferred to USC)
GPA: 3.9/4.0B.S. in Computer Science
Taiwan (Completed B.S. CS)
GPA: 3.8/4.0