Andrew Tulloch (Thinking Machines): 10 Key Things You Must Know

Overview

Andrew Tulloch is a distinguished Australian computer scientist and machine learning expert, best known as the co-founder of Thinking Machines Lab alongside Mira Murati, the former Chief Technology Officer of OpenAI. With a strong academic pedigree from the University of Sydney and the University of Cambridge, Tulloch has played a pivotal role in advancing AI technologies, particularly in large-scale machine learning systems. His career spans major tech powerhouses including Meta (Facebook) and OpenAI, where he contributed significantly to AI frameworks and models like PyTorch and GPT-4. Tulloch’s decision to reject a meteoric $1.5 billion offer from Meta to remain focused on his AI startup highlights his commitment to innovation and independence in the rapidly evolving AI landscape. This article explores ten fascinating aspects of Andrew Tulloch’s career, contributions, and impact.

1. Academic and Early Career Foundations

Andrew Tulloch began his academic journey in Perth, Australia, excelling at the University of Sydney where he earned a medal in mathematics. He furthered his expertise at the University of Cambridge’s Trinity College, graduating with distinction in mathematical statistics and machine learning, a testament to his deep analytical skills. Early in his career, he worked at Goldman Sachs as a quantitative analyst, leveraging his quantitative abilities in finance before transitioning fully into AI research. His strong foundation in mathematics and statistics has underpinned his impactful contributions in machine learning.

2. Pioneering Work at Meta (Facebook)

Tulloch spent over a decade at Meta (formerly Facebook), focusing on machine learning infrastructure critical for scaling AI applications. He was instrumental in developing PyTorch, one of the most widely adopted machine learning frameworks globally, which revolutionized AI research and deployment. His contributions at Meta spanned building distributed machine learning systems that supported real-time advertising applications and experimentation with large-scale models. This work established him as a significant figure in AI engineering and infrastructure.

3. Contributions to OpenAI and GPT-4 Development

After his tenure at Meta, Tulloch joined OpenAI, where he contributed to pretraining and improving models including GPT-4, one of the most advanced large language models. His role was pivotal in enhancing reasoning and training architectures that underpin modern AI capabilities. His time at OpenAI bridged fundamental AI research with practical, scalable implementations, helping to propel the organization’s AI models to new frontiers of performance and reliability.

4. Co-Founding Thinking Machines Lab

In early 2025, Tulloch co-founded Thinking Machines Lab with Mira Murati. The startup is focused on developing AI systems that go beyond traditional chatbot paradigms, emphasizing safety, interpretability, and customization. Despite being a relatively young company, Thinking Machines has quickly gained stature in the AI ecosystem and secured substantial investment, including a $12 billion valuation in its seed round. The company aims to tackle some of the most challenging problems in AI alignment and control.

5. Rejecting a $1.5 Billion Offer from Meta

One of the most talked-about chapters in Tulloch's career is his decision to decline an extraordinary $1.5 billion offer over six years from Meta, designed to entice him away from his startup. This offer, reportedly including bonuses and stock incentives, was part of Meta CEO Mark Zuckerberg’s aggressive strategy to acquire Thinking Machines Lab or at least hire its top talent amid intensifying competition in AI. Tulloch’s rejection symbolizes a notable shift in Silicon Valley, where purpose and innovation can outweigh even the most lucrative financial incentives.

6. Vision for AI Beyond Chatbots

Tulloch’s work with Thinking Machines reflects his vision to move AI beyond the current chatbot-centric user interfaces. The startup focuses on building AI tools that are safer, more interpretable, and customizable for complex real-world tasks, addressing key concerns about AI governance and usability. This approach aims to help usher in the next phase of AI where models are not just conversational but are deeply integrated and responsibly managed across various industries.

7. Impact on AI Recruitment Competition

Tulloch has been at the center of a fierce talent competition in AI, highlighted by Meta’s failed attempts to acquire Thinking Machines Lab and recruit its employees. His departure from Thinking Machines to Meta in late 2025, after initially resisting overtures, underscores the fluid dynamics of AI talent acquisition. Companies like Meta, OpenAI, and startups continuously vie for top researchers, reflecting Tulloch’s status as a prized expert and the high stakes in the AI innovation race.

8. Technical Proficiency and Programming Expertise

Beyond leadership and vision, Tulloch is known for his strong technical skills, proficient in multiple programming languages including C++, Python, Go, and Rust. His expertise encompasses developing large-scale machine learning systems and distributed computing algorithms—skills that are crucial for building scalable AI platforms. His technical versatility enables him to contribute hands-on to core AI infrastructure development as well as high-level architectural design.

9. Recognition and Influence in AI Circles

Although maintaining a relatively low public profile, Tulloch is widely respected in AI research and engineering communities. His contributions to foundational tools like PyTorch have had a profound influence on how AI models are built and deployed worldwide. His academic research in statistical learning theory and practical industry impact position him as a key figure bridging theoretical and applied AI.

10. Future Prospects and AI Industry Influence

Looking ahead, Andrew Tulloch and Thinking Machines Lab stand to shape the future trajectory of AI development through their focus on safe, interpretable, and highly capable AI systems. Their early success in fundraising and talent acquisition signals potential to significantly influence AI architectures and applications globally. Tulloch's career choices reflect a broader trend of innovator-driven startups challenging incumbent tech giants in a high-stakes AI innovation landscape.

Conclusion

Andrew Tulloch’s journey from a high-achieving mathematics student in Australia to a key architect of transformative AI technologies encapsulates the dynamic evolution of the artificial intelligence field. His work at Meta, OpenAI, and as co-founder of Thinking Machines Lab demonstrates a blend of intellectual rigor, technical mastery, and visionary leadership. By turning down lucrative offers to prioritize innovation and independence, Tulloch exemplifies a new generation of AI pioneers shaping the future of technology. His story invites reflection on the values driving AI’s next frontier—will the future be defined by bold startups and visionary engineers like Tulloch, or by the powerful tech giants seeking to dominate the landscape?