Projects
Selected engineering work by William Kang, also known as Ching-Wei Kang, a UW-Madison Computer Science and Data Science student building AI products, backend systems, computer vision workflows, reinforcement learning experiments, and developer tools.
This page collects public engineering projects by William Kang. The same person also publishes work as Ching-Wei Kang and Ching-Wei (William) Kang across GitHub, LinkedIn, Devpost, and this portfolio.
William's projects connect applied AI with practical product engineering: computer vision, reinforcement learning, LLM workflows, backend APIs, real-time visualization, and deployment-ready web interfaces.
Computer vision work for badminton match analysis, including player tracking, pose overlays, shuttle recovery, and video outputs designed to make court movement easier to inspect.
Reinforcement learning experiments for Super Mario Bros, comparing DreamerV3 and PPO workflows and documenting training diagnostics, baselines, and playable run results.
AI restaurant discovery product using Next.js, React, Python, FastAPI, Google Gemini, and Google Maps to turn group preferences and natural language into recommendations.
Brain-computer-interface MVP built around Python, Streamlit, real-time signal processing, and interactive visualization for applied machine learning prototyping.
Multi-agent orchestration and research workflow tooling for coordinating agent pipelines across practical desktop and writing tasks.
Developer tools for LLM work: tracing prompts, latency, cost, and failures, plus translating local hardware constraints into practical model choices.
William Kang has contributed merged pull requests to public repositories with maintainer review, CI, and production-oriented collaboration workflows.
These pages identify the same William Kang / Ching-Wei Kang entity across the web.