| layout | home |
|---|---|
| author_profile | true |
Ph.D. Candidate - U of M-Dearborn | Applied AI Researcher in Reliable Deep Perception
Research Vision: Building hybrid perception systems that know when predictions break physics, and know when to abstain, not just when to be confident.
Industrial & Systems Engineering
Geometry-Informed Computer Vision
Hybrid Machine Learning
I am a Ph.D. Candidate in Industrial and Systems Engineering at the University of Michigan-Dearborn. My dissertation, Examining Transportation Safety and Road User Behaviors through Machine Learning and Vision Transformers,
develops data-driven pipelines that integrate vehicle kinematics, projective geometry, and vision models to analyze driver behavior and vehicle–bicyclist interactions, with an eye toward systems humans can understand, oversee, and trust in safety-critical settings.
<p>My work with both data-driven and physics-informed approaches has revealed limitations in each method individually. This informs my future research on <strong>Limit-Aware Hybrid AI</strong> frameworks organized around three core thrusts: <strong>physically grounded architectures</strong> that embed domain constraints as structural components, <strong>geometry-informed representation learning</strong> that leverages mathematical models of space and motion, and <strong>operational awareness with structured abstention</strong> that enables systems to know when to defer rather than confidently predict. These frameworks create perception systems that are both statistically accurate and operationally robust for safety-critical deployment. I am committed to <strong>open science</strong> and <strong>reproducible research</strong>. Currently seeking <strong>academic research positions</strong>.</p>
<h2 id="research-focus">Research Focus</h2>
<p>My current research develops <strong>geometry-informed computer vision and machine learning methods</strong> for <strong>transportation safety</strong> and <strong>driver behavior analysis</strong>. My future research vision focuses on <strong>Limit-Aware Hybrid AI</strong> for reliable deep perception, developing frameworks that embed physical and regulatory limits as structural components of perception architectures to create systems that are both statistically accurate and operationally robust.</p>
<h2>Education</h2>
<div class="education-list">
<div class="education-item">
<div class="degree">Ph.D. in Industrial and Systems Engineering</div>
<div class="institution">University of Michigan-Dearborn and Rackham Graduate School</div>
<div class="year">Aug 2026 (expected)</div>
</div>
<div class="education-item">
<div class="degree">M.S. in Human Centered Design and Engineering</div>
<div class="institution">University of Michigan-Dearborn</div>
<div class="year">Aug 2021</div>
</div>
<div class="education-item">
<div class="degree">B.E. in Computer Science and Engineering</div>
<div class="institution">Anna University</div>
<div class="year">Apr 2013</div>
</div>
</div>
<h2>Certifications</h2>
<div class="certification-list">
<div class="cert-item">
<div class="cert-name">Connected and Automated Transportation Certificate</div>
<div class="cert-issuer">Center for Connected and Automated Transportation</div>
<div class="cert-date">Apr 2026 (expected)</div>
</div>
<div class="cert-item">
<div class="cert-name">Rackham Professional Development Diversity, Equity, and Inclusion Certificate</div>
<div class="cert-issuer">Rackham Graduate School - University of Michigan</div>
<div class="cert-date">2025</div>
</div>
<div class="cert-item">
<div class="cert-name">Post Graduate Diploma in Computer Applications</div>
<div class="cert-issuer">Computer Software Research Institution, India</div>
<div class="cert-date">Dec 2011</div>
</div>
</div>
<h2>Research Interests</h2>
<div class="interests-list">
<ul>
<li class="interest-item"><strong>Methodological Focus</strong>: Geometry-informed computer vision and perception - Physics-informed and hybrid machine learning - Calibration-free sensing and projective geometry features - Physically grounded architectures for safety-critical systems - Geometry-informed representation learning - Operational awareness and structured abstention</li>
<li class="interest-item"><strong>Application Domains</strong>: Vulnerable road user safety - Cyclist and pedestrian protection - Transportation safety and naturalistic driving analysis - Reliable perception for autonomous and cyber-physical systems - Safety-critical decision support and human-interpretable uncertainty</li>
</ul>
</div>
<h2>Scholarships and Awards</h2>
<div class="awards">
<div class="award-item">
<div class="award-title">Student Visionary Award</div>
<div class="award-details">International Forum on Research Excellence (IFoRE' 25), Sigma Xi - The Scientific Research Honor Society</div>
<div class="award-info">2025</div>
</div>
<div class="award-item">
<div class="award-title">Upsilon Pi Epsilon (UPE) Scholarship</div>
<div class="award-details">Awarded for exceptional academic performance, extracurricular involvement, and leadership within the computing community</div>
<div class="award-info">2024</div>
</div>
<div class="award-item">
<div class="award-title">Global Finalist</div>
<div class="award-details">NASA Space Apps Challenge</div>
<div class="award-info">2023</div>
</div>
<div class="award-item">
<div class="award-title">Irma M. Wyman Scholar</div>
<div class="award-details">Center for the Education of Women (CEW+), University of Michigan</div>
<div class="award-info">$11,500 - 2020-2021</div>
</div>
<div class="award-item">
<div class="award-title">Non-Resident Graduate Student Scholar</div>
<div class="award-details">University of Michigan-Dearborn</div>
<div class="award-info">$13,000 - S2020, F2020, W2021</div>
</div>
<div class="award-item">
<div class="award-title">Deloitte Hackathon Special Mention; Syncfusion Hackathon 2nd Place</div>
<div class="award-details">Deloitte (2017); Syncfusion (2015, INR 35,000)</div>
<div class="award-info">2015–2017</div>
</div>
</div>