Tom Brown is the co-founder of Anthropic, a company focused on developing safe and aligned AI systems. With a background in computer science and engineering, Brown contributed to the AI landscape through his roles in AI startups and research initiatives. He played a role in engineering development, notably at OpenAI, before embarking on his journey with Anthropic. ^[1]​

Early Life and Education

Tom Brown was born and raised in Massachusetts, where he developed an early interest in technology. He pursued higher education at the Massachusetts Institute of Technology (MIT), from where he graduated with a Master of Engineering degree in 2010, focusing on computer and cognitive science. ^[1]

Career

Early Ventures

Tom Brown began his career in the startup sector, working at Lingt, a language-learning platform, and later co-founding Grouper, a group-dating application. His experience at Grouper led to professional connections within the technology industry, including Greg Brockman, who later introduced him to work at OpenAI.

Brown also worked as a Founding Engineer at MoPub. In that role, he developed and scaled the company's Python-based advertising server, increasing its capacity from approximately 20,000 monthly impressions to more than 1.5 billion. He also implemented caching systems to reduce latency, established interviewing and onboarding processes, and participated in recruiting six of the company's first 20 engineers.

From 2012 to 2015, Brown served as Co-Founder and Chief Technology Officer of Grouper, a Y Combinator-backed company that developed a platform for arranging social meetings between groups of friends.

OpenAI and DeepMind

Brown joined OpenAI in 2016 as a Member of Technical Staff, becoming one of the organization's early employees. During this period, he co-authored the research paper Deep RL from Human Preferences, which examined reinforcement learning from human feedback (RLHF).

Between 2017 and 2018, Brown was a Member of Technical Staff at Google DeepMind. Working with Ian Goodfellow's adversarial machine learning group, he contributed to research on attack and defense methods for machine learning systems.

Brown returned to OpenAI in 2018 as Research Engineering Lead for GPT-3. He led the engineering work for the model and was responsible for the training infrastructure used to scale it from 1.5 billion to 170 billion parameters. His work also included contributions to research on scaling laws for neural language models and the paper Language Models are Few-Shot Learners.

Anthropic

In 2021, Brown co-founded Anthropic with former OpenAI colleagues, including Dario Amodei and Daniela Amodei. The company was established as a public benefit corporation focused on artificial intelligence research and development. Brown serves as Co-Founder and Chief Compute Officer, overseeing the company's computing infrastructure and related technical operations.

Brown has participated in Anthropic's discussions with government agencies, including negotiations concerning export controls affecting the company's AI models.

In public interviews and discussions, Brown has described networking, self-directed learning, and mentorship as factors in his transition into artificial intelligence. He has also discussed the use of online educational resources and mentorship during his career development.

At Anthropic, Brown continues to oversee computing infrastructure and engineering activities related to the company's artificial intelligence research. ^{[1] [2] [4] [3] [5] [6] [7] [8]}