Phitchaya Mangpo Phothilimthana

I’m a Technical Member of Staff at OpenAI. I work on hardware-software co-design and build efficient software systems for evolving AI hardware. My work aims to make AI workloads faster, more power-efficient, and sustainable at scale.

Previously, I was a Research Scientist at Google DeepMind (formerly Google Brain), where I led the Machine Learning for Machine Learning Compilers initiative, one of the lab’s moonshot projects. I created the XLA TPU autotuner, which is deployed to optimize production AI workloads daily, and pioneered the use of AI to improve the XLA TPU compiler itself.

My research interests span compilers, AI for systems, systems for AI, and sustainable computing. I earned my Ph.D. in Computer Science from UC Berkeley and a Bachelor’s degree from MIT. I was a recipient of Microsoft Research PhD Fellowship, Qualcomm Innovation Fellowship, and Thai Government Scholarship.

Beyond systems research, I’m passionate about technology for social good. I’m the founder of Curico, a nonprofit organization dedicated to making curiosity-driven learning accessible to every child, through the power of AI. At UC Berkeley, I developed an automated hint-generation system for an introductory programming course. At MIT, I built a color-blindness accessibility tool and collaborated with an NGO in India to improve the efficiency of LED solar lanterns distributed to off-grid villages.