The Berkeley Artificial Intelligence Research Lab has graduated its 2026 class of Ph.D. students.
The group covers robotics, embodied intelligence, large language models, computer vision, generative modelling, AI safety, and human-AI interaction. Their output includes published research, systems with practical application, and mentorship within the department.
Graduates are now moving into faculty roles, postdoctoral positions, industry research labs, and their own startups. A number are currently deciding their next steps.
Graduates
Baifeng Shi
Advisor: Trevor Darrell
Focus: Generalist vision and robotic models.
Current role: Member of Technical Staff at Physical Intelligence.
Contact: ba*********@******ey.edu | Website
Charlie Snell
Advisor: Dan Klein
Focus: Understanding when and how different large language model scaling paradigms can be traded off and interchanged. The work addresses test-time scaling, which treats each prompt independently and discards inferences between prompts, contrasting this with pretraining that learns compressed representations from large datasets. The goal is to develop methods that convert test-time inferences into learned representations the model can retain across interactions.
Contact: cs******@******ey.edu | Website
Devin Guillory
Advisor: Trevor Darrell
Focus: Accounting for data shifts in computer vision models.
Current role: Building collaborative AI systems.
Contact: dg*******@******ey.edu | Website
Eve Fleisig
Advisor: Dan Klein
Focus: Designing language models to work reliably and fairly for a broad range of users. Research uses disagreement among user preferences as a signal to train and evaluate models for entire populations. Work also involves designing rigorous evaluations to identify harms faced by diverse users and addressing core technical failures, such as miscalibrated confidence, to reduce downstream risks when models are deployed to users with different needs. The combined interventions aim to build models that minimise societal harms and maximise benefits for real-world users.
Current role: Postdoctoral fellow at Princeton CITP.
Contact: ef******@******ey.edu | Website
Grace Luo
Advisor: Trevor Darrell
Focus: Interpreting and controlling generative models. Projects include re-purposing image generators for computer vision tasks and meta-modelling language activations for better large language model probing and steering.
Current role: Research scientist in industry.
Contact: gr******@******ey.edu | Website
Hanlin Zhu
Advisor: Stuart Russell, Jiantao Jiao
Focus: Understanding and improving the reasoning capabilities of large language models.
Current role: Member of Technical Staff at OpenAI.
Contact: ha*******@******ey.edu | Website
Haozhi Qi
Advisor: Jitendra Malik, Yi Ma
Focus: Dexterous Manipulation and Robot Learning.
Current role: Research scientist at Amazon and Faculty at University of Chicago.
Contact: hq*@******ey.edu | Website
J.D. Zamfirescu-Pereira
Advisor: Bjoern Hartmann
Focus: Effective human-AI co-design. The work studies the boundaries of language interfaces as a medium for interacting with AI, creating systems that blend language-focused interactions with structured user interfaces drawing on different levels of abstraction. It focuses on language-oriented technologies, such as large language models and text-to-image models, acting as powerful mediators of design processes. These technologies enable humans to describe desires at almost any level of abstraction, from high-level goals vaguely specified to low-level corrections of undesired outputs.
Contact: za***@******ey.edu | Website
