I built QwenPaw, an open-source Agent OS and AI assistant with multi-agent orchestration, a kernel-level sandbox, access policies, persistent memory, and integrations for Discord, Telegram, and Slack.
I also developed Codedog, an AI-powered code review tool that uses LLMs to generate pull request summaries, file-level feedback, automated reviews, notifications, and codebase health scores across GitHub and GitLab workflows.
At Maxgen Technologies, I ran 50+ model-based reinforcement learning simulations for robotic control and improved deep RL convergence by 25% using PyTorch, RLlib, and vectorized environments. I also bring cybersecurity experience assessing web vulnerabilities, including SQL injection and XSS.
