I am a Research Scientist at Google DeepMind in London, working on the Multi-Agent team.

I recently completed my Postdoc at Facebook Artificial Intelligence Research (FAIR), where I expanded my research to the field of Multi-Agent Reinforcement Learning. During my postdoc, I largely explored the problem space of multi-agent communication towards the goal of zero-shot communication, learning of protocols that can generalize to novel partners. My latest body of work was motivated by communication skills for embodied agent populations. As I move forward in my research journey, I find myself thinking more generally about social influence and learning within cooperative multi-agent systems. Particularly when agents must consider the collective interests of the population, in light of their own individual self-interests.

I completed my PhD in Computer Science in 2019 at the School of Interactive Computing at Georgia Institute of Technology (GT), where I was advised by Dr. Sonia Chernova, director of the Robot Autonomy and Interactive Learning Lab.  My thesis research lied in the space of Interactive Robot Learning, with a focus on Active Learning. It investigated the problem of enabling an interactive robotic agent with a probabilistic reasoning framework for inferring what queries to make and when to query, given the goal of learning new concepts from a human partner. Part of what made active learning in this context particularly interesting and challenging is the work assumed the setting of a non-stationary physical environment, where external time and budget constraints were being imposed on the agent. The learning scenario was inspired by more realistic human settings, where continual learning should occur and presumably the agent should be able to drive its own learning. 

My broader research interests span the space of autonomous reasoning and decision-making for artificial agents in multi-agent settings. To date, my work has largely focused on models and algorithms for enabling agents to learn through interaction with other agents (artificial or human), trading off cooperative learning objectives with environmentally-motivated constraints.  As my research career evolves, I aim to more deeply explore theoretical frameworks and methods for reasoning and learning within multi-agent problem settings.

News.

October 2022: I look forward to participating in the University College London (UCL) Black History Month Panel entitled: “Celebrating Black Computer Scientists”.

September 2022: I am serving on the AAAI-23 Organising Committee, as Diversity and Inclusion Co-Chair.

August 2022: I gave an Invited Talk as part of the Oxford Machine Learning (OxML 2022) Summer School.

June 2022: I served on the RSS-22 Organising Committee, as Co-Chair of the Inclusion Program.

May 2022: I gave an in-person :) Invited Talk in the Berkeley Simons Institute Workshop on Multi-Agent Reinforcement Learning and Bandit Learning. It was part of a series on Learning and Games. [video]

December 2021: I gave an Invited Talk in the COMARL Seminar Series! [video]

December 2021: I started as a Research Scientist at DeepMind, in London!

September 2021: I served on the Organising Committee for the NeurIPS 2021 Workshop on Cooperative AI, as Program Chair.

September 2021: I gave Invited Talks at two IROS 2021 Workshops: on Semantic Policy and Action Representations for Autonomous Robots and on Multi-Agent Interaction and Relational Reasoning [video].

August 2021: Excited to announce that I have accepted a Research Scientist position at DeepMind!

August 2021: After an amazing two-year journey, I completed my Postdoc at Facebook AI Research :)

June 2021: I gave an Invited Talk for a Microsoft Research Seminar Series, focusing on my Multi-Agent Communication work (same as bullet point below).

May 2021: I gave an Invited Talk at the Stanford NLP Seminar Series. The topic was my latest body of work on: Multi-Agent RL towards Zero-Shot Emergent Communication.

April 2021: I gave an Invited Talk at the Cornell Robotics Seminar Series. The talk spanned a larger body of my research (from both PhD and Postdoc) on: Learning through Interaction in Cooperative Multi-Agent Systems.

March 2021: We released an arXiv preprint of our latest work on Quasi-Equivalence Discovery for Zero-Shot Emergent Communication.

March 2021: I gave an Invited Talk for the DeepMind Language and Games Reading Group. The topic was my latest body of work on: Multi-Agent Reinforcement Learning towards Zero-Shot Emergent Communication.

February 2021: I gave an Invited Talk at the University of Colorado Boulder Applied Mathematics Colloquium.

December 2020: I presented a poster at the NeurIPS Deep Reinforcement Learning Workshop [video] and the NeurIPS Zero-Shot Emergent Communication Workshop. [poster]

November 2020: We released an arXiv preprint of our work on: Exploring Zero-Shot Emergent Communication in Embodied Multi-Agent Populations.

November 2020: I was accepted into EECS Rising Stars 2020! Hosted (virtually) by UC Berkeley this year.

October 2020: I gave an Invited Talk in the UMass Amherst Machine Learning and Friends Seminar Series. The topic was Learning through Interaction in Multi-Agent Systems (which can include human agents). [video]

October 2020: I gave a spotlight (lightning) talk at the University of Michigan 2020 AI Symposium.

July 2020: I gave an Invited Talk at the ICML workshop on Human in the Loop Learning. The topic was my thesis work on Active Learning in Realistic Human Settings. [video]

July 2020: I gave the Facebook Sponsor Talk at the ICML Women in Machine Learning (WiML) workshop. The topic was my newest work on Learning to Communicate Nonverbally for Embodied Agent Populations. [video] ==> more in-depth version of spotlight talk below (given at RSS)

June 2020: Our extended abstract on Emerging Nonverbal Communication Protocols was accepted into the RSS Workshop on Emergent Behaviors in Human-Robot Systems. [spotlight talk — 5 min]

December 2019: I got hooded at the GT doctoral graduation ceremony. I am officially a Doctor of Philosophy.

May 2019: I returned to GT and successfully defended my dissertation! Entitled: Managing Learning Interactions for Collaborative Robot Learning [phd defense talk]

May 2019: Our paper on Active Learning within Constrained Environments got accepted into IJCAI 2019.

April 2019: I started as a Postdoctoral Researcher at Facebook AI Research, in Pittsburgh.

February 2019: I was hosted as a guest on the GT School of Interactive Computing Interaction Hour Podcast series, with Professor and Chair Ayanna Howard.

* All paper reference information (and pdfs) listed under Publications tab.