Yearly, the Berkeley Synthetic Intelligence Analysis (BAIR) Lab produces among the most proficient and revolutionary minds in synthetic intelligence and machine studying. Our Ph.D. Graduates have every pushed the frontiers of synthetic intelligence analysis and are actually able to embark on new adventures in academia, trade, and different fields.
These unimaginable individuals carry a wealth of data, recent concepts, and the drive to proceed contributing to the development of synthetic intelligence. Their work at BAIR spans from deep studying, robotics, pure language processing to laptop imaginative and prescient, safety, and extra, making vital contributions to their respective fields and having a transformative impression on society.
This website is devoted to showcasing our colleagues and making it simpler for tutorial establishments, analysis organizations, and trade leaders to find and recruit the newest technology of synthetic intelligence pioneers. Right here you could find detailed biographies, analysis pursuits and get in touch with data for every of our graduates. We invite you to discover the potential collaborations and alternatives offered by these graduates as they search to use their experience and insights in new environments.
Be part of us in celebrating the achievements of BAIR’s latest PhD graduates. Their journey has simply begun and the long run they may assist construct is vibrant!
Due to our pals on the Stanford Synthetic Intelligence Laboratory for developing with this concept!
e-mail: salam_azad@berkeley.edu
web site: https://www.azadsalam.org/
guide: Ion Stoica
Analysis introduction: My analysis pursuits lie broadly within the fields of machine studying and synthetic intelligence. Throughout my PhD, I centered on surroundings technology/curriculum studying strategies for coaching autonomous brokers by means of reinforcement studying. Particularly, I’m dedicated to researching strategies to generate totally different coaching environments (i.e., studying eventualities) for autonomous brokers by means of algorithms to enhance generalization capabilities and pattern effectivity. Presently, I’m engaged on autonomous brokers based mostly on Giant Language Fashions (LLM).
Positions of curiosity: Analysis Scientist, Machine Studying Engineer
e-mail: aliciatsai@berkeley.edu
web site: https://www.aliciatsai.com/
guide: Laurent El Garvey
Analysis introduction: My analysis delves into the theoretical facets of deep implicit fashions, beginning with unified “state house” representations of simplified symbols. Moreover, my work explores varied coaching challenges related to deep studying, together with issues amenable to convex and non-convex optimization. Along with theoretical exploration, my analysis extends potential functions to a wide range of drawback domains, together with pure language processing and the pure sciences.
Positions of curiosity: Analysis Scientist, Utility Scientist, Machine Studying Engineer
e-mail: katherine22@berkeley.edu
web site: https://cwj22.github.io
guide: Tomitsuka Masairo, Wei Zhan
Analysis introduction: My analysis focuses on machine studying and management algorithms to deal with the difficult duties of autonomous racing in Gran Turismo Sport. I draw on my background in mechanical engineering to discover how machine studying and model-based optimum management can create secure, high-performance management methods for robotics and autonomous methods. I significantly emphasize how you can leverage offline datasets (resembling human participant racing trajectories) to offer higher, extra sample-efficient management algorithms.
Positions of curiosity: Analysis scientists and robotics/management engineers
e-mail: chawin.sitawarin@gmail.com
web site: https://chawins.github.io/
guide: David Wagner
Analysis introduction: I am very within the security and safety facets of machine studying methods. Most of my earlier work has been within the space of adversarial machine studying, particularly adversarial paradigms and the robustness of machine studying algorithms. Recently, I have been enthusiastic about rising safety and privateness dangers on giant language fashions.
Positions of curiosity: analysis scientist
e-mail: eko@berkeley.edu
web site: https://www.elizakosoy.com/
guide: Alison Gopnik
Analysis introduction: Eliza Kosoy works with professors on the intersection of kid growth and synthetic intelligence. Alison Gopnik. Her work contains creating evaluation benchmarks for the LL.M. rooted in little one growth, in addition to learning how kids and adults use GenAI fashions resembling ChatGPT/Dalle and type psychological fashions about them. She is an intern on Google’s AI/UX group and beforehand labored at Empathy Lab. She publishes in Neurips, ICML, ICLR, Cogsci, and Cognition. Her thesis work created a unified digital surroundings for testing kids and synthetic intelligence fashions in a single place to coach reinforcement studying fashions. She additionally has expertise founding startups and STEM hard-coded toys.
Positions of curiosity: Analysis Scientist (Baby Improvement and AI), AI Security (Makes a speciality of Kids), Person Expertise (UX) Researcher (Makes a speciality of Blended Strategies, Youth, AI, LLM), Schooling and AI (STEM Toys)
e-mail: fangyuwu@berkeley.edu
web site: https://fangyuwu.com/
guide: Alexander Bayan
Analysis introduction: Beneath the steering of Professor Alexandre Bayen of Fangyu, he focuses on the applying of optimization strategies in multi-agent robotic methods, particularly within the planning and management of autonomous autos.
Positions of curiosity: Trainer or analysis scientist in management, optimization, and robotics
e-mail: frances@berkeley.edu
web site: https://www.francesding.com/
guide: Jacob Steinhardt, Moritz Hart
Analysis introduction: My analysis focuses on machine studying for protein modeling. I work on enhancing protein property classification and protein design, in addition to understanding the educational content material of various protein fashions. I’ve beforehand labored on sequence fashions for DNA and RNA, in addition to benchmarks for evaluating the interpretability and equity of ML fashions throughout domains.
Positions of curiosity: analysis scientist
e-mail: kathyjang@gmail.com
web site: https://kathyjang.com
guide: Alexander Bayan
Analysis introduction: My thesis work specialised in reinforcement studying for autonomous autos, with a concentrate on enhancing decision-making and effectivity in utility contexts. In future work, I’m keen to use these rules to broader challenges throughout domains resembling pure language processing. With my background, I intention to see the direct impression of my efforts by contributing to revolutionary AI analysis and options.
Positions of curiosity: Machine Studying Analysis Scientist/Engineer
e-mail: nikhil_ghosh@berkeley.edu
web site: https://nikhil-ghosh-berkeley.github.io/
guide: Yu Bin, Track Mei
Analysis introduction: I’m fascinated by utilizing theoretical and empirical strategies to realize a greater elementary understanding of deep studying and enhance sensible methods. Presently, I’m significantly fascinated by enhancing the effectivity of huge fashions by learning how you can correctly scale hyperparameters based mostly on mannequin measurement.
Positions of curiosity: analysis scientist
e-mail: oliviawatkins@berkeley.edu
web site: https://aliengirlliv.github.io/oliviawatkins
guide: Peter Abell and Trevor Darrell
Analysis introduction: My work includes reinforcement studying, BC, studying from people, and agent studying utilizing widespread sense base mannequin reasoning. I am excited in regards to the studying, supervision, consistency, and robustness of language brokers.
Positions of curiosity: analysis scientist
e-mail: rcao@berkeley.edu
web site: https://rmcao.internet
guide: Laura Waller
Analysis introduction: My analysis pursuits are in computational imaging, particularly spatiotemporal modeling of dynamic scene restoration and movement estimation. I additionally work on mild microscopy strategies, optimization-based optical design, occasion digicam processing, and novel field-of-view rendering.
Positions of curiosity: Analysis scientist, postdoc, trainer
e-mail: ryanhoque@berkeley.edu
web site: https://ryanhoque.github.io
guide: Ken Goldberg
Analysis introduction: Imitation studying and reinforcement studying algorithms scale to giant fleets of robots performing operations and different complicated duties.
Positions of curiosity: analysis scientist
e-mail: sdt@berkeley.edu
web site: https://www.qxcv.internet/
guide: Stewart Russell
Analysis introduction: My analysis focuses on making language fashions secure, strong, and safe. I even have expertise with imaginative and prescient, planning, imitation studying, reinforcement studying, and reward studying.
Positions of curiosity: analysis scientist
e-mail: shishirpatil2007@gmail.com
web site: https://shishirpatil.github.io/
guide: Joseph Gonzalez
Analysis introduction: Gorilla LLM – Instruments for educating LLM (https://gorilla.cs.berkeley.edu/); LLM execution engine: guaranteeing the reversibility, robustness and minimization of explosion radius of LLM-Brokers and integrating it into customers and enterprises Workflow; POET: Reminiscence constraints and energy-saving fine-tuning of LLM on edge units resembling smartphones and laptops (https://poet.cs.berkeley.edu/).
Positions of curiosity: analysis scientist
e-mail: spetryk@berkeley.edu
web site: https://suziepetryk.com/
guide: Trevor Darrell, Joseph Gonzalez
Analysis introduction: I work to enhance the reliability and security of intermodal transportation fashions. My focus is on localizing and decreasing illusions in visible + language fashions, in addition to measuring and utilizing uncertainty and mitigating bias. My curiosity lies in making use of options to those challenges in actual manufacturing eventualities, not simply in tutorial settings.
Positions of curiosity: Generate utilized analysis scientists in synthetic intelligence, safety, and/or accessibility
e-mail: xingyu@berkeley.edu
web site: https://xingyu-lin.github.io/
guide: Peter Abel
Analysis introduction: My analysis course is robotics, machine studying and laptop imaginative and prescient. The principle purpose is to study common robotics expertise from two views: (1) Studying a structured world mannequin with spatial and temporal abstraction. (2) Pre-train visible representations and expertise to allow data switch from web-scale visible datasets and simulators.
Positions of curiosity: Trainer or Analysis Scientist
e-mail: yyu@eecs.berkeley.edu
web site: https://yaodongyu.github.io/
guide: Michael Jordan, Ma Yi
Analysis introduction: My analysis pursuits lie broadly within the concept and apply of reliable machine studying, together with explainability, privateness, and robustness.
Positions of curiosity: school