Info

Machine Learning Research Associate II

MBZUAI

Research on privacy risks of spiking neural networks against membership inference attacks. Contributed to torchgfn with graphical environment for GFlowNet analysis. Developed agentic framework for diverse melody generation via GFlowNet-finetuned LLMs. Focus on probabilistic generative modeling and cybersecurity applications.

• Worked on privacy risks of spiking neural networks against membership inference attacks
• Contributed to torchgfn with graphical environment for GFlowNet analysis
• Developed agentic framework for diverse melody generation via GFlowNet-finetuned LLMs
• Research focus on probabilistic generative modeling and cybersecurity applications

Machine Learning Research Associate

University of Waterloo AI Institute

Built adversarial test features for autonomous vehicle visual AI (AVL Scenius). Developed CARLA-based real-time object detection testing framework. Proposed a KDE-based classifier for adversarial vs regular vehicle identification. Collaborated with AVL on autonomous vehicle safety testing.

• Built adversarial test features for autonomous vehicle visual AI (AVL Scenius)
• Developed CARLA-based real-time object detection testing framework
• Proposed a KDE-based classifier for adversarial vs regular vehicle identification
• Collaborated with AVL on autonomous vehicle safety testing

Data Science Intern (MITACS Accelerate)

TrojAI Inc.

Surveyed 100+ physical adversarial threats and designed Multi-Patch attacks. Proposed model-agnostic defenses using scene total-variation metrics. Research on safeguarding vision models against physical adversaries. Thesis work on CNN robustness against physical corruptions.

• Surveyed 100+ physical adversarial threats and designed Multi-Patch attacks
• Proposed model-agnostic defenses using scene total-variation metrics
• Research on safeguarding vision models against physical adversaries
• Thesis work on CNN robustness against physical corruptions

Decision Analytics Associate

ZS Associates

Designed surveys and models for US-based medical equipment pricing. Built statistical models and performed business analytics in R/Python/Tableau. Developed pricing strategies for medical device manufacturers.

• Designed surveys and models for US-based medical equipment pricing
• Built statistical models and performed business analytics in R/Python/Tableau
• Developed pricing strategies for medical device manufacturers

PhD

MBZUAI

Machine Learning

• Thesis: Decision Making Under Uncertainty using Generative Flow Networks
• Supervisor: Dr. Salem Lahlou
• Research focus: GFlowNets, probabilistic modeling, cybersecurity

MSc

University of British Columbia

Computer Science

• Thesis: Towards Safeguarding CNNs Against Physical Corruptions
• Supervisor: Dr. Apurva Narayan, Dr. John Braun
• Research focus: Adversarial machine learning, computer vision security

B.Tech

College of Engineering, Pune

Instrumentation and Control Engineering

• Thesis: Hardware Implementation of Model Predictive Controller
• Supervisor: Dr. D. N. Sonawane
• Gold Medal for highest GPA in department
• Part of COEP-SAT-2 satellite initiative

MBZUAI PhD Scholarship

MBZUAI

Full PhD fellowship covering tuition and living expenses (AED 840K). Competitive scholarship for outstanding research potential.

MITACS Accelerate Award

MITACS

CA$50K scholarship for research on adversarial defenses. Industry-academia collaboration with TrojAI Inc.

MITACS Globalink Graduate Fellowship

MITACS

CA$15K for returning scholars for master's studies in Canada. Recognition for previous research excellence.

UBC Dean's Entrance Scholarship

University of British Columbia

CA$5K merit award for incoming graduate students. Based on academic excellence and research potential.

Gold Medalist

College of Engineering, Pune

Highest GPA in the department for B.Tech. (Instrumentation and Control Engineering). Recognition for academic excellence.

COEP Alumni Excellence Award

College of Engineering, Pune

INR 50K based on academic and extracurricular excellence. Comprehensive evaluation of student achievements.

MITACS Globalink Research Award

MITACS

CA$10K to conduct research at UBC as visiting scholar. International research experience opportunity.

Towards Safeguarding CNNs Against Physical Corruptions

UBC Master's Thesis

Comprehensive study of physical adversarial attacks on computer vision systems. Proposed novel defense mechanisms using scene total-variation metrics. Surveyed 100+ physical adversarial threats and attack methodologies.

Hardware Implementation of Model Predictive Controller

COEP Bachelor's Thesis

FPGA-based implementation of MPC for real-time control systems. Optimized control algorithms for embedded systems.