Three entry points for my work: portfolio, daily research signals, and agent ecosystem indexing.
🏠 Homepage • 📚 Papers Hub • 🤖 Agent Index
- 🔭 Researching computer vision, generative AI, video understanding & AI agent engineering
- 🌱 Exploring foundation models, diffusion models, Harness Engineering, MCP/A2A/ACP protocols
- 💬 Ask me about Python, PyTorch, video editing, arXiv paper tracking, Agent infrastructure
- ✍️ Author of Claw Runtime & Agent Ecosystem 2026, Harness Engineering, Harness POC, and MCP Deep Dive
| Project | Description | Topics |
|---|---|---|
| 📰 Daily Video Papers | Automated arXiv paper tracking hub for Video, World Models, Agents & Tone/Color research | arXiv video agents automation |
| 🤖 Awesome Agent Everything | Meta-index of AI Agent repositories — automated classification & tagging | agent MCP A2A awesome-list |
| 🔧 harness-poc | Minimal Agent Harness POC — Agent = Model + Harness (~300 lines Python) | agent harness LLM Python |
| 👗 cloth-match | Efficient cloth matching system with deep visual features | CV retrieval fashion |
| 🎯 multimodal | Multimodal learning research | multimodal vision language |
| 🚁 uav-dispatch | UAV dispatch and management platform | UAV dispatch platform |
📖 Explored / Tried (fork repos)
- VideoCoF — CVPR 2026 Highlight: Unified Video Editing with Temporal Reasoner
- LiveMoments — ICLR 2026: Key Photo Restoration in Live Photos
- DynVFX — SIGGRAPH Asia 2025: Augmenting Real Videos with Dynamic Content
- ReCo — CVPR: Region-Constrained In-Context Generation
- CVPR 2023 Challenge — CVPR 2023 Foundation Model Challenge, Track 2 solution
- claude-code — Claude Code runnable build (Bun + TypeScript)
- MimicMotion_train — MimicMotion training
- mmrotate — OpenMMLab Rotated Object Detection
The fourth column is the base model, the fifth column is the optimized model trained by me.
You are my — thank you! 😊