Mirah Shi
Hello! I'm a fifth-year PhD student in computer science at the University of Pennsylvania, where I'm fortunate to be advised by Michael Kearns and Aaron Roth.
My research interest is in building AI systems that can interact effectively with one another and with humans. I use theoretical and empirical methods to study multi-agent interactions and design systems that are more collaborative, aligned, and trustworthy. I am grateful to have received a Google PhD Fellowship for my research. My work has also been supported by AWS AI.
During my PhD, I've interned at Apple working on collaborative capabilities in LLMs and at Morgan Stanley working on game-theoretic models for trading. Before coming to Penn, I received my bachelors in math from Barnard College. I have also spent time at Pacific Northwest National Laboratory on the Mathematics of Data Science team. I pronounce my name my-ra shee.
Email:mirahshi at seas.upenn.edu
Recent News
-
Our paper on narrowing the collaboration gap accepted to NeurIPS 2026. See you in Sydney!
-
Starting as a research intern at Apple.
-
Our paper on emergent alignment accepted to ICML 2026. See you in Seoul!
-
Our paper on collaborative prediction accepted to SODA 2026.
-
Honored to be awarded the Google PhD Fellowship!
-
Spending the summer at Morgan Stanley ML Research.
Publications
(αβ) denotes alphabetical ordering. * denotes equal contribution.
-
Narrowing the Collaboration Gap, Probably
-
Personalization Aids Pluralistic Alignment under Competition
-
Emergent Alignment via Competition
-
Learning to Strategically Acquire Resources in Competition
-
Collaborative Prediction: Tractable Information Aggregation via Agreement
-
Algorithmic Aspects of Strategic Trading
-
Sample Efficient Omniprediction and Downstream Swap Regret for Non-Linear Losses
-
An Elementary Predictor Obtaining \(2\sqrt{T}+1\) Distance to Calibration
-
Forecasting for Swap Regret for All Downstream Agents
-
Filtering Higher-Order Datasets
-
Center-Embedding and Constituency in the Brain and a New Characterization of Context-Free Languages
Selected Talks
- Designing Collaborative Multi-Agent Systems: From Theory to PracticeApple MLR Reading Group
- Emergent (Pluralistic) Alignment via CompetitionICLR AIMS Workshop Best Paper Talk
- Collaborative Prediction: Tractable Information Aggregation via AgreementSODA
- Collaborative Prediction: Tractable Information Aggregation via AgreementNYU Student Theory Day
- Theoretical Perspectives on Human-Centered AIMorgan Stanley ML Seminar
- Sample Efficient Omniprediction and Downstream Swap Regret for Non-Linear LossesCOLT
- Forecasting for Swap Regret for All Downstream AgentsESIF Economics and AI+ML Meeting
- Forecasting for Swap Regret for All Downstream AgentsEC
Teaching
I've been a teaching assistant for the following courses at Penn:
and at Columbia:
- COMS 3261 Computer Science Theory (Spring 2021)