Recent projects:
Built 50+ AI products for startups — now shipping multi-agent + MCP-native systems that replace entire SaaS stacks in 2026
Recent projects:
- Multi-agent orchestration (LangGraph + CrewAI) running 24/7 in production — replaced 6 SaaS tools, $40K/mo saved
- MCP-native AI workflows wired into Slack/Notion/Postgres — Claude/GPT-5 agents acting on real company data
- Voice AI agents (Retell + Vapi + ElevenLabs v3) handling 15K+ calls/month with sub-300ms latency
- Agentic RAG legal research tool with guardian agents + LangSmith eval — 99.2% citation accuracy
- Agentic commerce storefront on Next.js 16 + AI SDK 5 — conversion up 38% via on-page agents
Stack: Next.js 16 + AI SDK 5 • React 19 + Compiler • TypeScript • Python • FastAPI • Supabase + pgvector • AWS / Vercel
Agents: LangGraph 1.0 • CrewAI • AutoGen / AG2 • OpenAI Agents SDK • DSPy • LlamaIndex • Inngest / Trigger.dev
Protocols: MCP (Model Context Protocol) • A2A (Agent-to-Agent) • Tool-calling • Structured outputs
Models: GPT-5 • Claude Opus 4.7 / Sonnet 4.6 • Gemini 2.5 • Llama 4 • on-device (WebGPU + Ollama)
Edge & Frontend: Server Components • Streaming UI • Astro islands • Qwik resumable • TanStack Start
Mobile: Kotlin Multiplatform + Compose • React Native New Architecture
Blockchain: Solana • Ethereum • Smart Contracts • on-chain agent wallets
Clients hire me to ship the same agentic systems Anthropic & OpenAI demo on stage — multi-agent teams that book meetings, write code, qualify leads, and run support without humans in the loop.
DM me what you're building — I'll map out exactly which agent framework + MCP servers give you a 10x edge in 2026.
Portfolio: https://chee-ming-site.vercel.app
WhatsApp: +1 (769) 352-4496
Recent projects:
- Multi-agent orchestration (LangGraph + CrewAI) running 24/7 in production — replaced 6 SaaS tools, $40K/mo saved
- MCP-native AI workflows wired into Slack/Notion/Postgres — Claude/GPT-5 agents acting on real company data
- Voice AI agents (Retell + Vapi + ElevenLabs v3) handling 15K+ calls/month with sub-300ms latency
- Agentic RAG legal research tool with guardian agents + LangSmith eval — 99.2% citation accuracy
- Agentic commerce storefront on Next.js 16 + AI SDK 5 — conversion up 38% via on-page agents
Stack: Next.js 16 + AI SDK 5 • React 19 + Compiler • TypeScript • Python • FastAPI • Supabase + pgvector • AWS / Vercel
Agents: LangGraph 1.0 • CrewAI • AutoGen / AG2 • OpenAI Agents SDK • DSPy • LlamaIndex • Inngest / Trigger.dev
Protocols: MCP (Model Context Protocol) • A2A (Agent-to-Agent) • Tool-calling • Structured outputs
Models: GPT-5 • Claude Opus 4.7 / Sonnet 4.6 • Gemini 2.5 • Llama 4 • on-device (WebGPU + Ollama)
Edge & Frontend: Server Components • Streaming UI • Astro islands • Qwik resumable • TanStack Start
Mobile: Kotlin Multiplatform + Compose • React Native New Architecture
Blockchain: Solana • Ethereum • Smart Contracts • on-chain agent wallets
Clients hire me to ship the same agentic systems Anthropic & OpenAI demo on stage — multi-agent teams that book meetings, write code, qualify leads, and run support without humans in the loop.
DM me what you're building — I'll map out exactly which agent framework + MCP servers give you a 10x edge in 2026.
Portfolio: https://chee-ming-site.vercel.app
WhatsApp: +1 (769) 352-4496
Custom AI/ML solutions built by Chee Ming.
