Profile · New York

NickGu

Do well. Do good.

Portrait of Nick Gu
Engineer, teacher, board member.

What I mean by good

Impact is not a niche. Technology does good when it returns time and self-awareness to people, instead of taking them. Doing good can also be good business.

Chapter one

Learn

Four schools, three continents.

Before agents, I studied the brain. My doctorate connected human memory and cognition with machine learning, through neuroimaging.

Degrees
2024 Ph.D., Computational Neuroscience LMU Munich Dissertation: Spatiotemporal Regularities in the Brain LMU Munich
2017 M.S., Computer Science Columbia University Columbia University
2015 B.S., Computer Science Illinois Institute of Technology Illinois Institute of Technology
2014 B.S., Intelligent Science & Technology Peking University Peking University

Chapter two

Create

Agents reliable enough to ship.

At Google, I lead the evaluation infrastructure that helps coding agents such as Gemini CLI improve. Before that, I built the test infrastructure for Android.

30% 85%

Task success of a browser agent, after a measurement-driven loop.

3 months 1 day

A cross-team Android migration, automated for 60+ teams.

100K+

Modules on the Android test-ownership service I built. It saves 3.2 engineer-years a year.

10k+

Agent evaluation tasks a day on BenchHub, the harness I built.

Google

Code Wiki

Documentation generated from code, so it stays current. I built its agentic code indexing and its semantic search.

codewiki.google ↗

Internal at Google

Legion

A tool for automated code review and project maintenance. It is similar in spirit to Sentry's Seer.

Compare: Sentry Seer ↗

Chapter three

Empower

Teaching is the multiplier.

I teach Agentic Engineering at Columbia. I taught 7,000 Google engineers. I also help lead communities that create opportunities for others.

Agentic Engineering. Lecture 01 in one minute. It covers what intelligence is, what an agent is, the parts of an AI agent, and how to build one. Lecture 01 ↗ Playlist ↗ Course site ↗

7,000

Google engineers taught

12,000 +

Learners in all

20,000 +

Members, Chinese Googler Network

8,000

Members, Pioneering Minds AI

Also serving

  • Board, Pioneering Minds AI
  • Global Lead, StratOps, Chinese Googler Network at Google
  • Non-profit Engagement Champion, Google NYC
  • Startup mentor and advisor, 2022–2024, Google for Startups
  • Career & Network Committee, Columbia CAA

Four books changed me.

PeopleMachines Think

Unwavering Focus

Dandapani

What to Think About Machines That Think

John Brockman, editor

Ernst Pöppel wrote an essay in it and co-edited the book below.

Do

Zeng Guofan: A Biography

Zhang Hongjie

Human Compatible

Stuart Russell

In print

  1. 2025 Principles for Building Planet-Scale Eco-Conscious Personal AI Assistants Chapter in Ecological Turn with Intelligence, edited by Maria Reinisch and Ernst Pöppel · World Scientific

Work with me

I am open to conversations about enterprise AI, responsible AI, and AI upskilling. Pick a topic and write to me. Your message goes straight to my inbox.

Topic

Teach

Advise

Your message goes straight to my inbox.