Ashwinee Panda

I am a 4th (final) year PhD student at Princeton University working with Prateek Mittal on trustworthy artificial intelligence and privacy preserving machine learning. I am funded by fellowships and grants , most recently the OpenAI Superalignment Fast Grant.

Starting September 2024 I will be a Postdoctoral Associate at UMD working with Tom Goldstein on large language models.

Before Princeton I worked in the UC Berkeley RISE Lab where I was co-advised by Joey Gonzalez and Raluca Ada Popa, researching federated learning.

If you are interested in working with me send me an email at [firstname]_at_[university]_dot_[edu].

CV  /  Google Scholar  /  Twitter  /  Github

Research

I am currently working on a number of topics in LLMs, including pretraining, alignment, safety and privacy. The title of my PhD thesis was "Challenges in Augmenting Large Language Models with Private Data".

We propose a new practical data extraction attack that we call "neural phishing". This attack enables an adversary to target and extract sensitive or personally identifiable information (PII), e.g., credit card numbers, from a model trained on user data.

We propose the first method for performing differentially private in-context learning. Our method generates sentences from in-context learning while keeping the in-context exemplars differentially private, that can be applied to blackbox APIs (ex RAG).

We propose the first method for generating visual adversarial examples that can serve as transferrable universal jailbreaks against aligned large language models.

We pretrain networks with synthetic images that have strong performance on downstream private computer vision tasks.

We propose the first method for performing differentially private fine-tuning of large language models without backpropagation. Our method is the first to provide a nontrivial privacy-utility tradeoff under pure differential privacy.

We generate differentially private images from non-privately trained diffusion models by analyzing the inherent privacy of stochastic sampling.

We propose a new hyperparameter optimization method for differentially private machine learning that massively reduces privacy costs and compute costs.

Neurotoxin is a novel model poisoning attack for federated learning that stays present in the system for up to 5X longer than the baseline attack.

SparseFed is a provably robust defense against model poisoning attacks in federated learning that uses server-side sparsification to avoid updating malicious neurons.

FetchSGD is a communication-efficient federated learning algorithm that compresses gradient updates with sketches.

Not Research

WeChat  /  LinkedIn  /  Instagram  /  Yelp  /  Goodreads /  Spotify

I was born and raised in San Jose, California. In high school I taught math, played sax, argued vociferously, sang, danced, and wrote slam poetry. Before studying EECS at Cal I spent the summer in China working at a robotics company. I've been back a couple times.

While at Berkeley I founded DiscreetAI, a venture-backed startup building privacy-preserving machine learning as-a-service. You can check out our ProductHunt launch or our GitHub for more information. Among other things we won the first YCombinator Hackathon and built federated learning solutions for Fortune 500 companies.

Website template from Jon Barron.