Hi, I'm Saket S. Chaturvedi

I am a Ph.D. candidate in Computer Engineering at Clemson University, working under the mentorship of Dr. Xiaoyong (Brian) Yuan. My research focuses on autonomous driving, retrieval-augmented generation (RAG), and healthcare AI, where I develop and deploy machine learning models to address real-world challenges. With 6+ years of experience in Deep Learning and 3+ years in AI Security, I have worked extensively on multimodal fusion, object detection, classification, language models, and AI security. During my Ph.D., I also gained hands-on experience with Camera and LiDAR sensors, including configuring and deploying camera-LiDAR systems on autonomous vehicles. In my free time, I enjoy playing cricket, listening to music, and reading books.

EDUCATION & TRAINING

Ph.D. in Computer Engineering

Clemson University, USA, 2024 – Present.

Ph.D. in Computer Science (Transferred)

Michigan Technological University, USA, 2021 – 2023.

M.S. in Computer Science

Michigan Technological University, USA, 2021 – 2023.

B.E. in Computer Science & Engineering

RTM Nagpur University, India, 2013 – 2017.

IntereSts

Autonomous Driving & Perception Systems
Multimodal Fusion
3D Object Detection & Classification
AI Security & Backdoor Attacks
Retrieval-Augmented Generation (RAG) & LLMs
Machine Learning for Healthcare AI

RESEARCH HIGHLIGHTS

BadFusion: 2D-Oriented Backdoor Attacks against 3D Object Detection

With the increasing adoption of camera-LiDAR fusion for high-fidelity 3D object detection, the robustness of these systems against adversarial threats is a growing concern.

We introduce BadFusion, an innovative 2D-oriented backdoor attack that preserves trigger effectiveness across the entire fusion process, achieving significantly higher attack success rates compared to existing 2D-oriented attacks on 3D object detection models.

Autonomous Driving in Adverse Weather

Despite advancements in deep neural networks, object detection in adverse weather remains a significant challenge due to sensor degradation under extreme conditions.

To address adverse weather detection challenges, our paper introduces a Global-Local Attention (GLA) framework that adaptively fuses multiple sensor streams—camera, gated, and LiDAR data—at two fusion stages:

Early-stage fusion via a local attention network, capturing localized feature variations.
Late-stage fusion via a global attention network, assigning higher importance to the most reliable modality under the current weather condition.