Yun-Tai Yang

Backend engineer working across distributed systems and applied AI.

SaaS BackendMicroservices and data pipelines on AWS, GCP, and Kubernetes, Open APIs for enterprise integrations, and OAuth-based identity and access control
Applied AIAgent harnesses that handle orchestration, constraints, and tool access, MCP servers that turn real APIs into agent tools, and AI-native workflows shipped to production

Work

TypeScript · npm

A human-in-the-loop harness framework for building a knowledge model with AI. Agents draft the model from source material and raise questions where evidence is unclear, people decide what is accepted, and every node cites its source. The ontology, sources, storage, agent, and views are all pluggable, which keeps the framework domain-agnostic. It also ships as a complete system with an agent runtime and a review studio.

built on braid

newledge

TypeScript

An AI-native learning tool for the AI era, built on braid. Rather than consuming every video, article, and podcast, the learner reviews the knowledge an AI distills from them and merges what is approved into a personal whiteboard.
underlies braid

openapi-mcp-gateway

Python · PyPI

Mounts multiple OpenAPI specs as MCP servers in one process, each with its own auth. Tool shaping rewrites names, parameters, and responses, dynamic exposure keeps large APIs within an agent’s context, and read-only endpoints are served as MCP resources.
- also embedded in
  
  conciergent
  
  Python · PyPI
  
  Turns any MCP server, or an OpenAPI spec through the embedded gateway, into a Slack, LINE, or Discord agent that executes real operations. OAuth is completed within the conversation, and tools marked destructive require confirmation before they run.

Experience

2020 – 2026

KDAN · Staff Backend Engineer

Led backend development for DottedSign, a SaaS e-signature platform.

Designed and operated microservices on AWS and Kubernetes, consolidating 10+ services under shared REST API guidelines
Brought AI capabilities into production, from LLM-based workflows to agents built on a shared harness

Education

2016 – 2020

National Cheng Kung University

M.S., Computer and Communication Engineering, in the Database and Software Engineering Lab, focused on software engineering and collaboration.

Thesis, Achieving Full Strength of Collaboration by Optimizing Team Developer Capability
Journal paper, A noise blocking methodology for effective collaboration, Software: Practice and Experience (2024)

2012 – 2016

National Cheng Kung University

B.S., Electrical Engineering.