At Ormi Labs, I led the frontend implementation of an AI Agent workflow that turns natural-language requests into deployment, configuration, and environment-management tasks. I used Claude Tool Use and multi-turn agentic loops to coordinate tool selection, parameter gathering, execution feedback, and error handling.
I designed a state-machine-based interaction framework for tool calls, plan previews, user confirmations, and results. I also developed a Node.js AI Gateway with SSE streaming and Langfuse tracing, connecting agent runs, model calls, and tool executions for troubleshooting and usage analysis.
Across my work at Ormi Labs and Huobi, I developed frontend architecture and platform capabilities for developer tools, multi-tenant permissions, and blockchain products. I also independently delivered two iOS games and a WeChat mini-game, from concept and gameplay development through testing and launch.
