Platform & AI Systems Engineer · Beijing

I build platforms other teams depend on — and make them trustworthy enough to depend on.

10+ years across Microsoft, Amazon and Hulu/Disney+. I work where product velocity, reliability and cost pull in different directions — large-scale migrations, AI agents with real quality gates, and systems that stay up at 3 a.m.

10+years building platforms
65KCPU cores removed in one migration
99.97%availability on the MSN platform
150+Sev2/3 incidents as on-call DRI
01 — Geography

Journey

Jinan → Beijing → Stony Brook → Seattle → Los Angeles → Beijing. Rotate the globe, select a city, and replay how the work changed.

chapterroute
drag to rotate · scroll to zoom · click a city
Flight Log06
02 — Track record

Experience

Five companies, one theme: platforms that other teams build on.

03 — Proof

Case studies

Four systems, four different kinds of hard.

Rewriting the plane mid-flight

Windows 11 Start, taskbar and Edge new-tab feeds migrated while the downstream kept growing.

CPU cores
−65K
VMs
−1,700
P99 latency
−35.7%
Availability
99.97%
~$4M
saved per year · zero user-visible outage
contract→rollout order→rollback

One gateway, seven teams

A shared API gateway built from zero — standardised auth, versioning and rollout.

Before
300ms
After
3ms
7
partner teams adopted it
contract→versioning→canary→rollback

Status cascades, delivery follows

Ad lifecycle as a state machine: one bad transition at the asset level breaks delivery at the campaign level.

01Asset
02Creative
03Line item
04Campaign
ingest→validate→publish→deliver
$20M+
interactive ad revenue · 50M+ subscribers launched

Making “ship it” measurable

An agent platform is only useful if its output can be trusted — and trust needs gates.

plan→tools→groundedness→regression gate→guardrails
20 marketsdaily refreshbad-case feedback
20
markets running the agentic platform
04 — Toolkit

Skills

What I reach for, and what I've actually operated.

Systems
Operator
10+ YEARS
05 — Foundations

Education

Education flight path Bachelor's and master's at USTB in Beijing, followed by a master's in computer science at Stony Brook University in New York. USTB · B.S. INFORMATION & COMPUTATION BEIJING / 2006–2010 USTB · M.Eng. COMPUTER SCIENCE BEIJING / 2010–2013 STONY BROOK · M.S. COMPUTER SCIENCE NEW YORK / 2014–2015 ACADEMIC TRAJECTORY / LONGITUDE −73.12° → 116.40°
2006 — 2010
University of Science and Technology BeijingB.S., Information and Computational Science
2010 — 2013
University of Science and Technology BeijingM.Eng., Computer Science and Technology
2014 — 2015
Stony Brook UniversityM.S., Computer Science · New York