Azalia Mirhoseini
Azalia Mirhoseini is a computer scientist who applies machine learning to language models, computing systems and chip design. Her work spans mixture-of-experts models, learned device placement and AlphaChip, a reinforcement-learning system for chip layouts. She co-authored Constitutional AI and received two research awards in 2025. As of October 2026, she directs Stanford’s Scaling Intelligence lab and is a co-founder and CTO of Ricursive Intelligence.
Timeline
-
2015
Completing doctoral research
Her doctoral research is in electrical and computer engineering at Rice University. The ECE department gives her thesis its Best Thesis Award in 2015.
-
2017
Scaling networks, placing computation
She co-authors a sparsely gated mixture-of-experts layer for language modeling and machine translation, with model architectures of up to 137 billion parameters. In another paper, she trains a recurrent network to place neural computations on devices; the method outperforms handcrafted placement heuristics.
-
2019
Building AI for chip design
She is a research scientist at Google Brain, where she applies AI to chip design. MIT Technology Review names her to its Innovators Under 35 list.
-
2020
Learning a chip floorplan
She and Anna Goldie release a preprint on a reinforcement-learning method for chip layouts.
-
2022
Training assistants by principles
She co-authors Constitutional AI: Harmlessness from AI Feedback, which trains a harmless assistant through self-improvement guided by rules or principles. Her industry work includes roles at Anthropic, Google Brain and Google DeepMind.
-
2024
Giving a chip system a name
She and Anna Goldie announce the name AlphaChip for their chip-layout system. Google DeepMind says it has generated layouts used in every generation of Google’s TPU since its publication in 2020.
-
2025
Awards for self-improving AI
She receives the Okawa Research Grant Award for Self-Improving AI Systems and the Google ML and Systems Junior Faculty Award.
-
As of October 2026
Leading research at Stanford
She is an Assistant Professor of Computer Science at Stanford University, where she directs Scaling Intelligence. She is also a co-founder and CTO of Ricursive Intelligence, which develops AI tools for chip design.
Sources and credits
Photo: azaliamirhoseini.com; Azalia Mirhosesini; creator not stated. License: Not stated by source. Source: azaliamirhoseini.com. Cropped to a square and resized. Ask us to remove or replace this photo.
- Seminar @ Cornell Tech: Azalia Mirhoseini
- Azalia Mirhoseini
- Device Placement Optimization with Reinforcement Learning
- Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
- Azalia Mirhoseini | Innovators Under 35
- How AlphaChip transformed computer chip design
- Constitutional AI: Harmlessness from AI Feedback
- Azalia Mirhoseini | Stanford University School of Engineering
- Can AI Make AI Smarter? — TITV [Video] — The Information