Developer with 5+ years of experience, focused on AI applications and agent engineering, with strong React / Next.js foundations and Node.js, Python / FastAPI full-stack capabilities.
I'm Penn (Chen Peng), a developer with 5+ years of experience, currently targeting AI application, agent engineering, and full-stack roles where frontend depth matters.
For AI applications, I independently built AgentPay with multi-step tool calling, dynamic tools, human-in-the-loop authorization, conversational memory, RAG, read-only MCP, Agent Eval, and Run / Step Trace. It won runner-up in the Cobo hackathon track.
My core stack includes React / Next.js, TypeScript, Vue 3, Node.js, Supabase, and PostgreSQL. I use Python, FastAPI, and LangGraph to build the agent core service and controllable state graphs.
My earlier work includes Shenzhen Metro's ticketing administration platform, a digital-human metaverse product, and bidirectional IoT video. These projects built my experience in complex asynchronous state, cross-platform workflows, troubleshooting, and end-to-end delivery.
AI Agent Engineering
Tool orchestration / controlled execution
Connecting natural-language intent to tools, authorization, memory, RAG, evaluation, and tracing with explicit model and business boundaries.
Full-Stack Applications
Next.js / APIs / data
Building frontend, authentication, APIs, persistence, agent services, and deployment with Next.js, Node.js, Python / FastAPI, Supabase, and PostgreSQL.
Frontend & Cross-Platform
React / Vue / multi-platform delivery
Hands-on with React, Vue 3, TypeScript, UniApp, Flutter, and WeChat Mini Programs for complex interfaces and asynchronous state.
Secure Web3 Payments
Pacts / idempotency / on-chain execution
Using Cobo Agentic Wallet, EVM, and contract interactions with owner approval, policy-scoped keys, and stable request identifiers around financial actions.
Live, verifiable work across AI agents, full-stack products, complex business tools, and secure Web3 payments.
AgentPay
A controllable AI payment agent combining LangGraph orchestration, human approval, idempotent orders, cited RAG, MCP tools, and policy-bounded on-chain execution.
NeWare
A multilingual product and documentation site for battery-testing workflows, including product discovery and administration flows.
Web3 Interview Memo
A structured personal learning library for Web3 and frontend interview practice, with question banks and project-based notes.
Career Pilot
A multi-user career operating system that turns resumes, project notes, and target roles into personalized daily plans, interview practice, and application tracking.
Build logs, architecture decisions, and technical retrospectives across AI agents, full-stack engineering, frontend state, Web3 security, and product delivery.
Live sources from the last 48 hours across headlines, markets, Web3, and tech culture. No stale fallback stories — if sources fail, the page says so.
Hiring for an AI application, agent engineering, full-stack, or frontend role—or interested in a product collaboration? Reach out below. The form opens your local mail client.
The badge loads lazily from LinkedIn. If it is blocked on your network, use the profile link above.
For roles, engineering discussions, or partnerships, reach me through WeChat, LinkedIn, X, or email.