Rudan Xiao · AI Engineer, Full-Stack Builder

01 // sys.info

agent.handoff()

nameRudan Xiao

roleAI Engineer · Full-Stack

orgHybridity AB

clearancePhD Computer Science

status ACTIVE

ai.stack Agentic orchestration, Multi-agent systems, RAG pipelines, MCP integrations, Knowledge graphs, LLM evaluation, Medical imaging AI

app.stack React, Next.js, TypeScript, Python, FastAPI, PostgreSQL, Neo4j, Vector stores, Docker, Kubernetes, GCP, CI/CD

I build AI systems and the products around them. Most of my work is agentic orchestration for regulated workflows: turning open-ended requests into structured, reviewable execution flows, with traceability, human oversight and evaluation built in. I also ship the rest of the stack that makes those systems usable: React and TypeScript front ends, Python and FastAPI services, graph and vector stores, and the CI/CD and autoscaling that keeps them running. A PhD in computer science underneath, and a habit of taking things all the way to production.

02 // agents.list()

Running Agents

Hybridity Orchestrator ACTIVE

Building an agentic orchestration layer for compliance-focused AI workflows, translating conversational requests into structured execution across evidence review, requirement analysis, control extraction, and reporting. Designed multi-agent execution flows, MCP-style tool integrations, and durable state models enabling traceability, replay, and operator-governed review.

> translating conversational request into execution flow...

> loading knowledge graph: EU & Swedish regulations

> spawning sub-agents: [evidence_review, requirement_analysis, policy_review, control_extraction]

> RAG pipeline: OpenAI + Gemini, parallel execution

> handoff → operator approval queued, audit event logged

> run complete: traceability ✓ replay ✓ comparison ✓

Agentic AIMCPMulti-AgentRAGFastAPIGCP

See all projects

03 // journey.log()

Execution Timeline

2019

PhD, Computer Science

Université Côte d'Azur & INRIA, France. Developed interpretable ML pipelines combining handcrafted radiomics features with deep learning for kidney cancer classification. Built a multi-task CNN for semi-supervised segmentation and classification. Published at MICCAI and ERMIA.

2022

Karolinska Institute

Developed deep learning models in PyTorch for early detection and prognosis of breast cancer from 2D mammography and 3D MRI. Built multimodal AI systems integrating imaging data and clinical information for survival prediction.

2024-Present

Hybridity AB

Building an agentic orchestration layer for compliance-focused AI workflows, from the operator-facing interface down to the services behind it. Productionized LLM microservices with CI/CD and autoscaling. Built a knowledge-graph-based database covering EU and Swedish regulations. Developed end-to-end RAG and multi-agent pipelines with parallel and sequential execution patterns.

04 // tools.registry()