🚀 I'll be joining NVIDIA Research to work on AI Security and Privacy.
After almost 2 years at JP Morgan’s Explainable AI CoE, I’ve decided to take up a new role at NVIDIA Research, where I’ll be focusing on AI Security and Privacy.
About Me
I am a Sr. Research Scientist at NVIDIA. My research is focused on Trustworthy AI, with a particular emphasis on security and privacy.
I obtained my PhD from Georgia Tech in 2022, where I was advised by Prof. Moinuddin K. Qureshi. In addition to my work on Trustworthy AI, I have published in the areas of computer architecture, hardware security and AI accelerators.
I previously worked at JP Morgan’s Explainable AI CoE. I’ve also been fortunate to intern at several industrial research labs including FAIR and IBM Research during my PhD.
Interests
- AI Security & Privacy
- Trustworthy AI
Education
-
PhD
Georgia Tech
-
MS
Georgia Tech
-
BE
Sri Jayachamarajendra College of Engineering
News
🎉 Our work "Progressive Inference- Explaining Decoder-Only Sequence Classification Models Using Intermediate Predictions" was accepted at ICML 2024!
A new XAI framework to explain decoder-only sequence classification models.
🎉 Our work "Information Flow Control in Machine Learning through Modular Model Architecture" was accepted at Usenix Security 2024!
Enabling secure machine learning for access controlled data.
🎉 Our work "SHAP@k- Efficient and Probably Approximately Correct (PAC) Identification of Top-K Features" was accepted for an oral presentation at AAAI 2024!
Efficiently identifying important features with PAC guarantees.
Recent Publications
(2024).
Information flow control in machine learning through modular model architecture.
33rd USENIX Security Symposium (USENIX Security 24).
(2024).
Progressive Inference: Explaining Decoder-Only Sequence Classification Models Using Intermediate Predictions.
International Conference on Machine Learning.
(2024).
SHAP@ k: Efficient and Probably Approximately Correct (PAC) Identification of Top-k Features.
Proceedings of the AAAI Conference on Artificial Intelligence.
(2023).
Bounding the invertibility of privacy-preserving instance encoding using fisher information.
Advances in Neural Information Processing Systems.
(2023).
Cocktail party attack: Breaking aggregation-based privacy in federated learning using independent component analysis.
International Conference on Machine Learning.
Experience
-
Sr. Research Scientist
NVIDIAAI Privacy and Security -
Sr. AI Research Associate
JP Morgan Chase, XAI CoEExplainable AI -
AI Research Intern
Meta, FAIRFederated Learning -
AI Research Intern
Meta AIUncertainty quantification, Conformal prediction -
AI Research Intern
Meta AISemi-supervised learning to improve conversion prediction -
AI Research Intern
IBM ResearchNoise resilient deep learning models for analog AI accelerators
Education
-
PhD
Georgia TechRead ThesisGPA: 4.0/4.0
Thesis: Understanding and Mitigating Privacy Vulnerabilities in Deep Learning. Supervised by Prof. Moinuddin Qureshi.
-
MS
Georgia TechGPA: 4.0/4.0 -
BE
Sri Jayachamarajendra College of EngineeringGPA: 3.78/4.0