Roberta Raileanu

I am a Senior Staff Research Scientist at Google DeepMind where I lead the Open-Endedness team. I am also an Honorary Associate Professor (Adjunct Professor) at UCL, working closely with the UCL-DARK Lab where I advise PhD students and co-teach a new course on Open-Endedness and General Intelligence.

Previously, I was a Research Scientist at Meta GenAI where I led the AI Scientist team. Our goal was to accelerate AI Research and Development using AI Agents that can generate ideas and hypotheses, implement new methods, train ML models, run experiments, analyze results, and iterate through this process to make new scientific discoveries. At Meta, I also started and led the Tool Use team for Llama 3. This enabled Llama to use tools such as a search engine, Python interpreter, text-to-image models, and Wolfram Alpha, as well as zero-shot generalize to new tools at test time. Our research on LLM Agents enabled new products used my hundreds of millions of users such as Meta AI, Data Analyst, AI Studio, and the Ads Business Agent.

In 2021, I obtained my PhD in Computer Science from NYU, advised by Rob Fergus. My focus was on deep reinforcement learning. During my PhD, I was fortunate to intern at DeepMind, Microsoft Research, and Facebook AI Research.

Previously, I obtained a B.A. in Astrophysics from Princeton University, where I worked with Michael Strauss on theoretical cosmology and Eve Ostriker on supernovae simulations.

In a past life (aka high school), I was fortunate to have the opportunity of competing in IPhO, IOAA, IAO, as well as national math and computer science olympiads.

Email  /  CV  /  Google Scholar  /  GitHub  /  Twitter  /  LinkedIn

I am broadly interested in designing machine learning algorithms that can make robust sequential decisions in complex environments, while constantly acquiring new skills and knowledge. My long-term goal is to develop models that can solve a wide range of new tasks with little or no supervision and training. I also care deeply about ensuring such systems are reliable, trustworthy, and aligned with human intentions.

My work draws from fields such as reinforcement learning, open-ended learning, self-supervised learning, and natural language processing. Currently, I work at the intersection of large language models (LLMs) and decision making / RL. In particular, I focus on augmenting LLMs with decision making abilities (e.g. planning, reasoning, actions, tools, and goals), and teaching them to learn from (human) feedback and interaction in an open-ended learning fashion.

I'm also interested in understanding and improving the generalization and robustness of autonomous agents and large pretrained models. In the past, I've worked on deep reinforcement learning, exploration, fast adaptation, learning from demonstrations, as well as continual and multi-agent learning.

A new framework and benchmark to accelerate AI Research and Development using AI Agents.

Introduce MaestroMotif, a new method for training hierarchical LLM agents to solve complex compositional tasks requiring hundreds of steps.