Introduction

Hi! I am a PhD candidate in the Purdue University School of ECE, working with Professor Zahra Ghodsi in the Trustworthy and Collaborative Learning Systems (TCLS) lab.

Research

Research illustration (click me)

My work is strongly tied to Security and Privacy, Accelerator Architecture and Machine Learning. Specifically, I explore ways to make trustworthy ML more efficient by leveraging the compute capability of GPUs. My aim is to bridge the performance goals of the accelerator community with the security concerns of the ML community, where tasks are increasingly distributed across parties and new attack vectors against ML are always discovered.

I have designed GPU accelerated frameworks for privacy-preserving computation, proposed end-to-end frameworks for ML artifact authentication, and achieved significant speedup in ML training with differential privacy. I am also involved in AI/ML security projects by the Open Source Security Foundation (OpenSSF), part of the Linux Foundation.

Bio

Born and raised in Malaysia, I did the first two years of undergrad at Taylor’s University, Malaysia and completed the rest at Purdue University in West Lafayette, graduating with a BSc. in Computer Engineering in 2020. After working in Chicago as a software test automation engineer for Amount (now FIS Amount), I went to graduate school at Purdue and received a MSc. in ECE in 2024 before continuing to PhD candidacy. I aspire to join the industry as a researcher and transform ideas into innovations that will benefit many.

Publications

One RNG to Rule Them All - How Randomness Becomes an Attack Vector in Machine Learning

- Kotekar Annapoorna Prabhu, Andrew Gan, Zahra Ghodsi
- IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)
- March 2026
- Paper, Proceedings

Sentry: Authenticating Machine Learning Artifacts on the Fly

- Andrew Gan, Zahra Ghodsi
- ACM SIGSAC Conference on Computer and Communications Security (CCS)
- October 2025
- Paper, Slides, Proceedings

cuOT: Accelerating Oblivious Transfer on GPUs for Privacy-preserving Computation

- Andrew Gan, Setsuna Yuki, Timothy Rogers, Zahra Ghodsi
- IEEE International Symposium on Hardware Oriented Security and Trust (HOST)
- May 2025
- Paper, Slides, Proceedings

Projects

GPU-Based Model Integrity SIG

- A project by the OpenSSF AI/ML Security workgroup, part of the Linux Foundation
- Establish hardware-agnostic workflow for GPU-based model authentication
- Model trainers produce cryptographically signed artifacts with accelerated integrity operations

Teaching

ECE 47920: Privacy-preserving Machine Learning [Spring 2025]
ECE 36200: Microprocessor Systems and Interfacing [Spring 2023]
ECE 39595: Object-oriented Programming in C++ [Fall 2022]

Awards

Design Automation Conference (DAC) 2025 Young Fellow