Senior Staff Research Scientist / Meta TBD Lab

Zhaodong Wang

I build production AI agents, ML systems, and research-to-product loops for software that has to work every day.

AI agents / ML systems / product engineering / applied research

One paper / recent focus

Agent infrastructure, compressed.

The old site was a research archive. This version is intentionally narrower: recent agent systems first, research as proof, education as context.

Ship
Production agent systems

Build AI agents for real work surfaces, where reliability, tool use, and trust matter more than demos.

Build
Product + infrastructure

Own the connective tissue across product UX, orchestration, memory, tools, evaluation, and debugging.

Loop
Research to product

Turn frontier model behavior into measured software systems through shipping, observation, and iteration.

What I do now

I make agents operational.

Agent platforms

I build the product surface and the system underneath it.

Harnesses, orchestration, tools, memory, evaluation loops, debugging surfaces, and the product patterns that make agents usable by real people.

Model-product loops

I turn frontier behavior into reliable software.

The work is a tight loop between modeling, product, infrastructure, and user friction: ship, observe, evaluate, and feed the hard cases back into the system.

Systems taste

I care about reliability, legibility, and the last mile.

My background spans ML systems, distributed infrastructure, traffic engineering, research prototypes, and production product engineering.

Selected systems

Recent work that defines the profile.

Meta TBD Lab / Agentic AI

Production agent platforms

Led product and systems work on production AI agents, spanning user surfaces, harnesses, tools, and evaluation loops.

Product + Modeling

Agent harnesses and eval loops

Co-designing the scaffolding that lets agents use tools, remember context, recover from mistakes, and improve through measured feedback.

Meta / Netflix / Research

ML systems and network intelligence

Built prediction, optimization, automation, and applied ML systems for infrastructure at production scale.

Research signal

Research is present, not the whole story.

Background

A compact arc.

2022 - Present

Meta TBD Lab

Senior Staff Research Scientist leading agentic AI products and systems.

2021 - 2022

Netflix

Senior Research Engineer on ML systems and production infrastructure.

Education

UIUC + Tsinghua

Ph.D. Transportation, M.S. Applied Mathematics, and B.E. Civil Engineering and Economics.

Contact

Building agent systems and AI products.