NLP & language technology
Representation, retrieval, and evaluation for text-heavy products—where “good enough in the lab” is not good enough in production.
// Gordon Yeh
Mechanical engineer by training, NLP scientist by practice — Kaohsiung, Shanghai, Boston, two years in Palo Alto at Shein, and now fully remote from Long Island City, New York.
NLP scientist · Kaohsiung → Shanghai → Boston → Palo Alto (Shein) → Long Island City, NYC — mostly remote, always tied to real product stakes.
I was born March 30, 1995 in Kaohsiung, Taiwan. I grew up around industry—my parents run an air compressor business there—so early on I cared how machines, reliability, and operations actually behave, not just how they look on a slide.
At 18 I left for China for my bachelor’s at Shanghai Jiao Tong University, in the UM–SJTU Joint Institute with the University of Michigan, majoring in mechanical engineering. That training still shapes how I think about structure, constraints, and systems that have to work under load.
Later I moved to the U.S. for a master’s in data analytics engineering at Northeastern University in Boston, then stepped into analytics and ML in industry—from a McKinsey co-op through NLP and search/recommendation roles—always trying to connect models to something measurable on the business side.
Today I live in Long Island City, Queens, New York, and work fully remote. At Empath (also remote) I was based in Boston. At Shein I spent two years on site in Palo Alto before this coast-to-coast thread landed me in New York.
Fully remote from Long Island City, New York. Building and shipping NLP for products that have to work outside the notebook.
Based in Palo Alto — ranking, retrieval, and relevance at scale for e-commerce discovery.
Fully remote while living in Boston — first full-time industry role applying NLP end to end.
Internship / co-op exposure to analytics and client-facing problem solving.
M.S. in Data Analytics Engineering — statistics, ML pipelines, and turning data into decisions.
B.S. in Mechanical Engineering (joint program with the University of Michigan). Foundation in dynamics, design, and systems thinking.
How you see markets, teams, risk, or time. This is not a résumé — it is the lens you use when trade-offs get real.
“Drop in a principle you actually live by — one sentence that people remember.”
Representation, retrieval, and evaluation for text-heavy products—where “good enough in the lab” is not good enough in production.
Ranking, query understanding, and feedback loops from Shein-scale discovery problems.
Formal ME training (UM–SJTU JI / SJTU) as a bias toward constraints, failure modes, and systems that have to run continuously.
Short notes, hypotheses, or patterns you have noticed. Link out to longer writing when you add it.