PeopleofSuperintelligence

Researcher

Javad Lavaei

Javad Lavaei is an associate professor at UC Berkeley whose AI work includes safe reinforcement learning and language-model reasoning. He earned engineering degrees at Sharif University of Technology in 2003 and Concordia University in 2007, then completed a control and dynamical systems PhD at Caltech in 2011. After postdoctoral work at Stanford, he was an assistant professor of electrical engineering at Columbia from 2012 to 2015 and joined Berkeley in 2015. He received the Presidential Early Career Award for Scientists and Engineers in 2019 and became an IEEE Fellow in 2022. In 2025, he coauthored DRCORL, a method for learning policies from fixed data under safety constraints, and StyleBench, a benchmark for comparing how language models reason.

Works
280
Citations
6,754
h-index
32

Publication numbers from OpenAlex, as of 28 October 2026.

Timeline

  1. 2003–2007

    Degrees at Sharif and Concordia

    He earns a BSc in electronics engineering at Sharif University of Technology in 2003. In 2007, he completes an MASc in electrical engineering at Concordia University.

  2. 2011–2015

    Caltech, Stanford and Columbia

    He completes a PhD in control and dynamical systems at Caltech in 2011 under John C. Doyle and Richard Murray. He spends one year as a postdoctoral scholar in Electrical Engineering and the Precourt Institute for Energy at Stanford University. From 2012 to 2015, he serves as an assistant professor in Electrical Engineering at Columbia University.

  3. 2015

    Joining Berkeley IEOR

    In 2015, he joins UC Berkeley's Department of Industrial Engineering and Operations Research. His work develops optimization algorithms for generating and delivering electricity and control methods for power grids with renewable energy.

  4. 2015–2017

    Power flow and optimization prizes

    His paper with Steven Low, “Zero Duality Gap in Optimal Power Flow Problem,” receives the PSACE Technical Committee Prize Paper Award in 2015. He also wins the INFORMS Optimization Society Prize for Young Researchers in 2015. The American Automatic Control Council gives him the Donald P. Eckman Award in 2016, and he receives the SIAM Control and Systems Theory Prize in 2017.

  5. 2019

    A presidential early career award

    In 2019, he receives the Presidential Early Career Award for Scientists and Engineers after nomination by the US Department of Defense. His work applies control theory and optimization to energy and efficient power distribution.

  6. 2022

    Fellowships and the Ruberti Prize

    In 2022, he becomes a Fellow of IEEE and the Asia-Pacific Artificial Intelligence Association. He also receives the Control Systems Society’s Antonio Ruberti Young Researcher Prize in 2022.

  7. 2024

    Pausing reinforcement learning updates

    With Hyunin Lee, Ming Jin and Somayeh Sojoudi, he coauthors “Pausing Policy Learning in Non-stationary Reinforcement Learning” at ICML in 2024. The work studies when to pause policy updates as an environment changes and computes an optimal ratio between policy update and hold duration.

  8. 2025

    DRCORL for safe learning

    In 2025, he coauthors the DRCORL method for offline reinforcement learning under safety constraints. It uses a diffusion model to learn the behaviour policy from a fixed dataset, then extracts a simpler policy for efficient use and adjusts gradients to balance rewards and safety constraints.

  9. 2025

    StyleBench tests reasoning styles

    He coauthors StyleBench in 2025 as a senior author with Junyu Guo and collaborators. The benchmark compares five reasoning styles across five task types and 15 open-source language models, examining how the choice of reasoning method affects accuracy and computing cost.

  10. As of October 2026

    Associate professor at Berkeley

    As of October 2026, he is an associate professor in UC Berkeley's Department of Industrial Engineering and Operations Research. His 2026 work includes neural-network architectures and failures in language models when their context changes, alongside control theory, optimization and power systems.

Career

  1. NowUniversity of California, BerkeleyAssociate professor
  2. From 2015University of California, Berkeley
  3. 2012–2015Columbia UniversityAssistant professor
  4. Stanford UniversityPostdoctoral scholar

Education

  1. 2011California Institute of TechnologyPhD in control and dynamical systems
  2. 2007Concordia UniversityMASc in electrical engineering
  3. 2003Sharif University of TechnologyBSc in electronics engineering

From the biography.

Checked career sources: lavaei.ieor.berkeley.edu.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Javad Lavaei yet. Send us a free-licensed photo.

  1. Biography lavaei.ieor.berkeley.edu. Checked 28 October 2026.
  2. Javad Lavaei Joins Berkeley IEOR ieor.berkeley.edu. Checked 28 October 2026.
  3. Professor Low and Dr. Lavaei Receive PSACE Prize Paper Award eas.caltech.edu. Checked 28 October 2026.
  4. Selected Awards and Honors lavaei.ieor.berkeley.edu. Checked 28 October 2026.
  5. Javad Lavaei Wins 2016 Donald P. Eckman Award ieor.berkeley.edu. Checked 28 October 2026.
  6. White House honors engineering faculty with early career awards engineering.berkeley.edu. Checked 28 October 2026.
  7. Pausing Policy Learning in Non-stationary Reinforcement Learning proceedings.mlr.press. Checked 28 October 2026.
  8. Don't Trade Off Safety: Diffusion Regularization for Constrained Offline RL arxiv.org. Checked 28 October 2026.
  9. StyleBench: Evaluating thinking styles in Large Language Models arxiv.org. Checked 28 October 2026.
  10. How AI prompting shapes accuracy, efficiency and cost ieor.berkeley.edu. Checked 28 October 2026.
  11. Laboratory for Control, Optimization, and Power lavaei.ieor.berkeley.edu. Checked 28 October 2026.