Andrew Tulloch Joined Anthropic Inference Team
The veteran researcher moves to the San Francisco startup to focus on training Claude technology.
Updated on Sept. 28, 2026 in Artificial Intelligence
Live Poll
Do you trust the rapid development of AI by competing tech firms?
Andrew Tulloch has joined the inference and performance team at Anthropic as of September 2026. He transitions to the role following an 11-month tenure at Meta.
Why it matters
Tulloch brings over a decade of experience in AI research to Anthropic, where his focus will center on scaling and optimizing the underlying architecture of the Claude model family.
Tulloch joins Anthropic with 11 years of prior experience in Meta's AI research department. This follows a 12-month period at OpenAI during 2023 and 2024.
The players
Andrew Tulloch
A researcher and co-founder of the AI start-up Thinking Machines Lab with a decade-long career in artificial intelligence.
Anthropic
A San Francisco-based AI company that develops the Claude series of large language models.
Meta
A global technology firm focused on social platforms, AR/VR hardware, and large-scale AI research.
OpenAI
A research organization and company known for developing the GPT series of large language models.
The details
Tulloch joins the inference and performance team, the group responsible for optimizing how models like Claude process data during inference—the phase where a trained model executes tasks and generates outputs. His work will focus on training infrastructure, which involves the specialized software and hardware systems required to teach large-scale neural networks to process information.
Timeline
2007: Andrew Tulloch graduated from Christ Church Grammar School.
2023-2024: Tulloch worked for OpenAI for 12 months.
August 2025: The Wall Street Journal reported on a Meta offer.
September 2026: Andrew Tulloch moved to Anthropic.
The Tech Race
Tulloch's transition highlights the ongoing movement of veteran researchers between the major labs driving the current generative AI cycle. His addition to Anthropic follows his tenure at Meta, which concurrently launched its Muse AI agent in September 2026.
Readers who utilize the Claude platform may see performance improvements in model inference speeds as Tulloch’s team implements new training optimizations. There are no immediate changes to existing subscriptions or user workflows resulting from this personnel move.
The takeaway
Tulloch's move reflects the high demand for experts capable of refining inference performance in competitive AI development environments. Watch for upcoming performance benchmarks from the Claude model family to gauge the impact of recent adjustments to the underlying inference stack.
Further reading
For broader trends in research talent and scaling, visit our Artificial Intelligence section.
Source note: This article includes information reported by News.
Live Poll
Do you trust the rapid development of AI by competing tech firms?
