Who is Uday Ruddarraju? The engineer building the computing backbone behind OpenAI’s AI ambitions

Artificial intelligence is often judged by the capabilities of its models, but developing and running them also requires enormous computing resources. That is the engineering challenge Uday Ruddarraju has spent much of his career working on. Now OpenAI’s chief technology officer for compute, Ruddarraju has moved from building cloud infrastructure at eBay to leading infrastructure teams at Robinhood and Elon Musk’s xAI.

His experience offers a glimpse into the less visible work behind advanced AI: connecting computing hardware, networks, storage and data centres so researchers can train increasingly capable models. In his recent LinkedIn posts, Ruddarraju has outlined why he believes computing capacity will become a critical constraint as AI agents spread into more areas of work and research.

MUST READ | OpenAI appoints new compute chief: Indian-origin Uday Ruddarraju takes charge of data centers and compute push

From cloud infrastructure to AI systems

Ruddarraju studied computer science at Hyderabad’s Chaitanya Bharathi Institute of Technology before earning a master’s degree in the subject from the University of Minnesota in the US.

His professional career began with early internships, including at Amazon, before he joined eBay in 2013. Over more than five years, he progressed through engineering roles focused on cloud infrastructure and platforms. He subsequently joined Robinhood, where he rose to senior director of engineering and head of infrastructure.

DON'T MISS | The AI boom is hiding a bigger technology divide: Who profits as device sales fall?

In June 2024, he moved to xAI as head of infrastructure engineering, before joining OpenAI in July 2025. He became CTO of compute in July 2026, according to his LinkedIn profile.

Why compute matters to OpenAI’s next phase

In a LinkedIn post announcing his CTO role, Ruddarraju described work spanning computing, networking, storage and machine learning systems used to train frontier models such as GPT-5.6. He also highlighted the scale of the engineering effort ahead, including distributed systems, hardware manufacturing and data centre construction.

His more recent post explains how OpenAI intends to bring commercial strategy closer to engineering decisions. That includes working with infrastructure partners, developing its own chips and building computing capacity directly.

Ruddarraju also pointed to an internal frontier model’s mathematical results as an example of what additional compute could make possible. He said the average result used computing equivalent to roughly three hours of ChatGPT Pro thinking.

For Ruddarraju, the challenge is not simply building more data centres. It is ensuring the computing infrastructure can keep pace with the needs of AI research and the applications that emerge from it.