w.winhok
中文

AI products,
from interface to execution.

winhok · AI Full-Stack & Agent Engineer

I connect usable interfaces, agent runtimes and knowledge systems. Explore how the work behaves, the decisions behind it, and the evidence you can inspect.

Explore selected work About & contact

React / TypeScript · Python / Node.js · Agents / RAG

Finahead product walkthrough showing a financial decision
Production products in motion.Interview · interview practice / Finahead · financial decisionsWatch product walkthroughs →

Selected work

Two product journeys, a public runtime, and agent orchestration in motion.

All case studies

Knolume

AI products

Interview and Finahead, connected by a shared platform without merging their data.

A private multi-product AI platform powering Interview and Finahead, with shared account access and agent execution while keeping product data and workflows isolated.

Watch two product walkthroughs and read how the products keep their workflows and data separate.

  • React
  • TypeScript
  • NestJS
  • PostgreSQL
Read case study Explore InterviewExplore Finahead

Knolume Runtime

Agent systems

Reusable agent execution without baking product policy into the library.

A public TypeScript library for agent runs, tools, context, memory, retrieval, guardrails and sub-agents. Host applications provide models, tools, identity and product policy.

Inspect public source, a deterministic example and the versioned v0.4.0 release.

  • TypeScript
  • AI SDK
  • Tool calling
  • Guardrails
Read case study Source

Agent OS

Agent systems

Keeping agent work moving across tools, people and sessions.

A persistent orchestration layer for Claude Code and Codex. Coordinate agents, workspaces, approvals and scheduled execution from Feishu / Lark.

Watch an 84-second recorded task, from dispatch to a follow-up in the same native session.

  • TypeScript
  • Claude Code
  • Codex
  • MCP
Read case study Watch demoSource

The decisions are part of the work.

Explore my approach

Start with a real workflow.Connect the interface, data and model behavior around a task someone can complete.

Make failure a visible state.Plan for interrupted sessions, denied tools and unsupported answers.

Keep the proof close.Show the implementation, an inspectable demo and the limits of each result.

Building an AI team?

My focus is AI full-stack and Agent engineering. Start with a case study, or get in touch.

Get in touch