(*) Equal contribution

AI Safety & LLM Alignment

I study how harmful or hidden behaviors can emerge from realistic LLM training pipelines and how to mitigate them. This includes targeted manipulation induced by optimization for user feedback, as well as inference-time consensus methods that suppress behaviors introduced by poisoned or otherwise problematic fine-tuning data without requiring labels for safe source models.

Inference-Time Consensus for Mitigating Hidden Behaviors from LLM Fine-Tuning Adhyyan Narang*, Artin Tajdini*, Claire Zhang, Jamie Morgenstern. Accepted to the AdvML-Frontiers × CoTMA Workshop at COLM 2026 (paper)

On Targeted Manipulation and Deception when Optimizing LLMs for User Feedback Marcus Williams*, Micah Carroll*, Adhyyan Narang, Constantin Weisser, Brendan Murphy, Anca Dragan. ICLR 2025 (arxiv)

Feedback Loops & Multi-Agent Learning

When ML models are deployed in the real world, they shape the data they are trained on and the behavior of other agents. A recommender system influences what users consume; a pricing algorithm changes competitor behavior; a language model shapes the distribution of text it will later be fine-tuned on. I develop game-theoretic and optimization frameworks for understanding and controlling these feedback dynamics, building on tools from performative prediction, statistical learning and dynamical systems.

Dynamics of Learning under User Choice: Overspecialization and Peer-Model Probing Adhyyan Narang, Sarah Dean, Lillian J Ratliff, Maryam Fazel. UAI 2026 (arxiv)

Multiplayer Performative Prediction: Learning in Decision-Dependent Games Adhyyan Narang, Evan Faulkner, Dmitriy Drusvyatskiy, Maryam Fazel, Lillian J Ratliff. JMLR 2023 (arxiv)

Learning in Stochastic Monotone Games with Decision-Dependent Data Adhyyan Narang, Evan Faulkner, Dmitriy Drusvyatskiy, Maryam Fazel, Lillian J Ratliff. AISTATS 2022 (PMLR)

Global Convergence to Local Minmax Equilibrium in Classes of Nonconvex Zero-Sum Games Tanner Fiez, Lillian J. Ratliff, Eric Mazumdar, Evan Faulkner, Adhyyan Narang. NeurIPS 2021 (NeurIPS)

Sample-Efficient Learning

How can agents learn to make good decisions with as few samples as possible? We study this question in two settings: reinforcement learning, where we show how to exploit structural similarities between policies to dramatically reduce sample complexity; and black-box optimization of submodular + supermodular objectives under bandit feedback, with applications to active learning, summarization, and content recommendation.

Sample Complexity Reduction via Policy Difference Estimation in Tabular Reinforcement Learning Adhyyan Narang, Andrew Wagenmaker, Lillian J. Ratliff, Kevin Jamieson. NeurIPS (Spotlight) 2024 (arxiv)

Efficient Interactive Maximization of BP and Weakly Submodular Objectives Adhyyan Narang, Omid Sadeghi, Lillian J Ratliff, Maryam Fazel, Jeff Bilmes. UAI 2024 (arxiv)

Foundations of Overparameterized Learning

Modern neural networks have far more parameters than training points, yet generalize well in practice — defying classical statistical theory. We study this phenomenon in linear models, showing that classification is fundamentally easier than regression in overparameterized settings, and that overparameterization can make models brittle to adversarial perturbations even when standard performance is good. We also study meta-learning under overparameterization, giving insight into where useful inductive biases come from.

Classification and Adversarial Examples in an Overparameterized Linear Model: A Signal Processing Perspective Adhyyan Narang, Vidya Muthukumar, Anant Sahai. ICML Workshop on Overparameterization 2021 (arxiv)

Towards Sample-Efficient Overparameterized Meta-Learning Yue Sun, Adhyyan Narang, Halil Ibrahim Gulluk, Samet Oymak, Maryam Fazel. NeurIPS 2021 (arxiv)

Classification vs. Regression in Overparameterized Regimes: Does the Loss Function Matter? Vidya Muthukumar*, Adhyyan Narang*, Vignesh Subramanian*, Mikhail Belkin, Daniel Hsu, Anant Sahai. JMLR 2021 (arxiv)