Dr. Manas Gaur

Assistant Professor · Tenure-Track

Department of Computer Science and Electrical Engineering

College of Engineering and Information Technology

He/Him/His/Himself

About

Manas Gaur is an Assistant Professor of Computer Science at UMBC and Director of the KAI² Lab, where he pursues use-inspired basic research at the intersection of neurosymbolic AI, mechanistic interpretability, and clinical natural language processing. His overarching mission is to render large language models trustworthy for high-stakes applications, particularly in mental health and clinical decision support. Gaur holds the prestigious Ramanujan Fellowship from the Anusandhan National Research Foundation, recognizing his standing among early-career researchers in artificial intelligence. He serves as an Advisor AI Research Scientist for NeuralNest and Aidvance. He brings industry-academic experience from Samsung Research America and Dataminr, combined with sustained clinical partnerships spanning the Maryland Psychiatric Research Center and university-embedded clinical psychology. His research traces a coherent arc from foundational work in social media mining for mental health, through the development of Knowledge-Infused Learning as a neurosymbolic paradigm, to mechanistic interpretability and grounding fidelity for clinical AI systems. His innovation portfolio includes three patents that define core algorithmic advances for high-stakes applications: AQGPT, ISEEQ, and Virtual Court Room, each anchoring trustworthy AI deployment in clinical and legal contexts. Gaur is deeply committed to early-stage researcher development and academic leadership. Through mentorship of undergraduates and high school students to first-author publication, he has seeded cumulative research impact now flowing through Carnegie Mellon, Purdue, and leading technology companies. His current research frames Knowledge-Infused Neurosymbolic AI around three scientific pillars, Interpret, Ground, and Control, positioning mechanistic reasoning as the foundation for trustworthy clinical AI.

Research interests

Neurosymbolic AI, Knowledge-Infused Learning, Mechanistic Interpretability, Sparse Autoencoders and Circuit Analysis, Long-Form Reasoning and Attribution, Retrieval-Augmented Generation, Grounding Fidelity in LLMs, Clinical Natural Language Processing, Trustworthy AI for High-Stakes Applications (Cybersecurity, Scientific Discovery, Mental Health, Legal), Adversarial Robustness

Teaching interests

Neurosymbolic AI Systems, Knowledge-infused Learning, Machine Learning, Semantic Mechanistic Interpretability, Trustworthy AI, Knowledge Graph Representation and AI Reasoning, Large Language Models, Natural Language Processing

Education

  • Ph D, Computer ScienceUniversity of South Carolina (2022)
    Knowledge-infused Learning
  • MS, Software EngineeringDelhi Technological University (Formerly Delhi College of Engineering) (2015)
    BIOGEOGRAPHY BASED OPTIMIZATION FOR COMPLEX SYSTEM
  • BS, Computer ScienceNetaji Subhas University of Technology (East Campus) (Formerly Ambedkar Institute of Technology) (2013)
    Meticulous study of firewall using security detection tools