Victoria Dean

Victoria Dean is a computer scientist whose work connects machine learning and robotics. At MIT, she founded and taught Code for Good and explored video prediction; at Deep Genomics, she built BRANCHR for RNA branch-point selection. At Waymo, she developed onboard text detection and CaRL, completed a robotics PhD at Carnegie Mellon, and joined Olin as an assistant professor. Her research includes See, Hear, Explore, TOTO, and Hearing Touch; in 2026, her team won first place at the SIGCSE ACM Student Research Competition for CubBot. Since 2025, she is a Lecturer in Computer Science at Princeton and runs the Smart Buildings Challenge at NeurIPS 2026.

Now at: Princeton University
Works: 14

Publication numbers from OpenAlex, as of 27 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

2014–2017

Code for Good, video and BRANCHR

At MIT, she earns a BS in Computer Science and develops her interest in teaching. From 2014 to 2017, she is instructor of record and founder of Code for Good, a course where students develop software for nonprofits. In the 2015–2016 SuperUROP cohort, she studies video prediction with deep convolutional networks; at Deep Genomics in 2016, she builds BRANCHR, a model for predicting branch-point selection in RNA splicing.
2017

Onboard text detection

At Waymo in 2017, she develops and deploys an onboard text-detection system. She uses an accurate but expensive Google OCR system to train a low-latency onboard model that helps cars recognize important words and disambiguate situations.
2018

CaRL at Waymo

In 2018, her CaRL work combines imitation learning with reinforcement learning for trajectory generation, with the goal of learning from observed behavior and generalizing beyond it.
2020

Audio-visual curiosity

She coauthors See, Hear, Explore in 2020, a curiosity method that rewards actions for novel associations between pixels and sounds. On Atari and AI Habitat, SHE allows more exploration, more sample-efficient learning, and greater robustness to noise than existing curiosity baselines.
2023

A robotics PhD and TOTO

She completes a PhD in Robotics at Carnegie Mellon's School of Computer Science in 2023 and joins Olin College of Engineering as an assistant professor. She coauthors Train Offline, Test Online (TOTO), a benchmark that gives remote users shared robots for evaluation and an open-source dataset for offline training.
2024

Hearing Touch

In 2024, she coauthors Hearing Touch, which uses contact microphones as an alternative tactile sensor and audio-visual pretraining to improve robotic manipulation.
2026

CubBot at SIGCSE

Her team wins first place at the 2026 SIGCSE ACM Student Research Competition for developing and evaluating CubBot, a custom LLM for COS 324 that studies generative AI's effects on student behavior and learning outcomes. She runs the Smart Buildings Challenge at NeurIPS 2026.
Since 2025

Lecturer at Princeton

She joins Princeton's Computer Science Department in 2025 after two years as an assistant professor at Olin College of Engineering and is a Lecturer in Computer Science. She teaches introductory computer science, machine learning, and ethics, with research interests in machine learning and robotics.