AI Application Developer ยท Building tools that turn ideas into reality
๐ง PH penguin โ I build AI agents that solve real problems, not just demo problems.
I believe great AI tools should deliver value to everyone โ no prompt engineering, no environment setup, just upload your data and get results. Every project I build starts from a real pain point:
- Tired of cleaning data? โ Built a one-click cleaning tool
- Agents have no memory? โ Created a local memory layer
- Deploying agents is painful? โ Made a one-click deployment platform
Motto:
Ship it, then polish.๐ขโจ
๐ค AgentFlow โ AI Agent Deployment Platform
Deploy agents as easily as Docker containers
|
Core Features
|
Technical Highlights
|
22โญ ยท ๐ฅ Most Popular Project
๐งช SmeltLabAI Data Cleaning + One-Click Modeling Upload dirty data โ Say one sentence โ Get clean data and a trained model
|
๐ง EngramUnified Memory Layer for Agents One memory for all your agents. Local, portable, MCP-native.
|
๐ AI Love AnalysisChat Subtext Analyzer for Couples Upload WeChat screenshots โ AI analyzes true intent
|
Languages
AI & Backend
Frontend
Infrastructure
I don't just build my own projects โ I also fix bugs in others ๐
- ๐ก๏ธ mem0 โ Security fixes, Docker hardening, Qdrant concurrent write protection
- ๐ open-webui โ DuckDuckGo search fix, SearXNG language handling
- ๐ธ browser-use โ Screenshot auto-compression for LLM context optimization
- ๐ deer-flow โ Event loop lifecycle fix in async memory updater
def my_contribution_philosophy():
return {
"focus": ["Security ๐", "Performance โก", "Bug fixes ๐", "DX improvements ๐ง"],
"approach": "Minimal, surgical patches โ fix the root cause, not the symptom",
"preference": "PRs > issues > comments"
}๐ AgentFlow v2 โ Advanced agent orchestration capabilities ๐งช SmeltLab โ Multi-class support and hyperparameter search ๐ง Engram โ MCP protocol support and cross-agent memory sharing ๐ Writing โ Sharing lessons learned along the way
