# Tom Brown

Co-founders · Chief Compute Officer at Anthropic · joined 2021

A startup engineer who taught himself AI research and went on to lead the engineering of GPT-3.

Full profile: https://www.autumn.ai/resources/anthropic/people/tom-brown

**At Anthropic:**
- Current: Co-founder and Chief Compute Officer
- Jan 2021: Co-founder
Sources: [1](https://www.anthropic.com/company/leadership) [2](https://www.linkedin.com/in/nottombrown)

**Career before Anthropic:**
- Dec 2018 – Dec 2020: OpenAI, Research Engineering Lead, GPT-3
- Aug 2017 – Nov 2018: Google DeepMind, Member of Technical Staff
- May 2016 – May 2017: OpenAI, Member of Technical Staff
- 2012–2015: Grouper, Co-Founder & CTO
- Feb 2011 – Nov 2011: MoPub, Founding Engineer
Sources: [1](https://www.linkedin.com/in/nottombrown)

**Education:**
- MEng, Course 6 (Computer Science), Course 9 (Brain and Cognitive Sciences), Massachusetts Institute of Technology, 2005-2010
Sources: [1](https://www.linkedin.com/in/nottombrown)

**Lesser-known facts:**
- Co-founded the startup Grouper and was a founding engineer at MoPub before AI. [1](https://www.linkedin.com/in/nottombrown)
- Got a B-minus in linear algebra, then taught himself with a Coursera course, Kaggle and the book Linear Algebra Done Right. [1](https://www.businessinsider.com/everything-we-know-about-tom-brown-anthropic-white-house-2026-7)
- Led research engineering for GPT-3 at OpenAI. [1](https://www.linkedin.com/in/nottombrown)
- Built MoPub's early server architecture and scaled its ad-serving API to 1.5 billion monthly impressions. [1](https://www.contextjamming.com/founder-files/tom-brown)
- Wrote the Grouper matching algorithms behind more than one million drinks across 25 cities before the startup shut down. [1](https://www.contextjamming.com/founder-files/tom-brown)
- Messaged Greg Brockman as soon as OpenAI was announced offering to help, and was eventually given a job on a StarCraft project. [1](https://www.businessinsider.com/everything-we-know-about-tom-brown-anthropic-white-house-2026-7)
- Wrote rl-teacher, open-source code for Deep RL from Human Preferences plus a web app for collecting human feedback. [1](https://github.com/nottombrown)
- Struck a 2026 deal to rent SpaceX compute, including the Colossus data center, for $1.25 billion a month. [1](https://www.contextjamming.com/founder-files/tom-brown)
- Gives this career advice: "Surround yourself with people you want to be like. You'll become more similar to them over time." [1](https://www.businessinsider.com/everything-we-know-about-tom-brown-anthropic-white-house-2026-7)

**Research papers before Anthropic:**
- Deep Reinforcement Learning from Human Preferences (2017), author 3 of 6: https://arxiv.org/abs/1706.03741
- Fine-Tuning Language Models from Human Preferences (2019), author 4 of 8: https://arxiv.org/abs/1909.08593
- Language Models are Few-Shot Learners (2020), author 1 of 31: https://arxiv.org/abs/2005.14165
- Scaling Laws for Autoregressive Generative Modeling (2020), author 8 of 19: https://arxiv.org/abs/2010.14701
- Scaling Laws for Neural Language Models (2020), author 4 of 10: https://arxiv.org/abs/2001.08361

**Interviews and talks:**
- [Building Anthropic: a conversation with the co-founders (2024)](https://www.youtube.com/watch?v=om2lIWXLLN4)

**Known for:**
- Deep RL from Human Preferences (2017)
- Language Models are Few-Shot Learners (GPT-3, 2020)
- Scaling Laws for Neural Language Models (2020)
Sources: [1](https://www.businessinsider.com/everything-we-know-about-tom-brown-anthropic-white-house-2026-7)

**Forbes estimate:** $15.5B as of 2026-10-18 (an estimate, not a disclosed holding). [1](https://www.forbes.com/profile/tom-brown/)

Researched with Autumn (https://www.autumn.ai) from public sources. Updated 2026-10-18. Everyone at Anthropic: https://www.autumn.ai/resources/anthropic/people