I build production-grade ML and backend systems (microservices, low-latency C++, full-stack apps) and research AI/ML reasoning and explainability in LLMs and knowledge graphs—bridging rigorous research and real-world engineering.
A Black Belt and Gold Medalist at the Wado International Karate Championship (2017), representing India. Years of training taught me precision, focus, and calm under pressure—qualities that mirror my approach to research and mentorship.
From being Class Representative to Junior Sports Leader, I learned that leadership is about empathy, accountability, and shared purpose. These lessons continue to shape how I guide research teams and mentor peers in academic and project settings.
With AIESEC Baroda's Outgoing Social Sector team, I coordinated cross-cultural exchange programs that fostered youth leadership and social impact—broadening my worldview and reinforcing my belief in globally responsible AI.
As a former Graduate Assistant at GWU, former Teaching Assistant at Navrachana University, and NVIDIA Jetson AI Project Coordinator, I've mentored over 100 students in AI/ML, big data & analytics, and software testing—guiding them in transforming theory into reproducible research, published work, and hands-on innovation.
I enjoy blending logic with creativity—whether designing tools, writing technical content, or crafting user-centric systems. Innovation, in my opinion, flourishes where structure meets imagination.
I recently completed my M.S. in Computer Science (Machine Intelligence & Cognition) at The George Washington University (GPA 3.88/4.0), where I served as a Graduate Assistant for Big Data & Analytics (CSCI 4907/6444). I work across two equally weighted tracks, engineering production-grade ML and backend systems and research on reasoning-centric AI systems, LLM alignment, and explainable AI, developing software that is reliable, scalable, and faithful in real-world settings.
On the engineering side, I design and ship full-stack, production-grade systems. My flagship project is the Multimodal AI Intelligence Platform (live at projectmmap.com), a deployed system for retrieval-augmented chat over text, PDFs, images, audio, and video, with a live knowledge graph and grounded, cited answers, built on FastAPI, async workers, Next.js, and a Qdrant/Neo4j/Postgres/Redis data plane. My other systems work includes TaskForge (microservices with gRPC), ResearchVault (a full-stack research repository), and OrderBook++ (a low-latency C++ matching engine).
On the research side, my recent work explores faithful reasoning and attribution grounding in language models. In RSAT, I train small LLMs to produce structured reasoning with cell-level citations, improving attribution faithfulness by 3.7×. In FASS (CVPR 2026 XAI4CV), I study the stability of explanations under realistic perturbations, highlighting critical gaps in current evaluation practices.
I also work on efficient and adaptive LLM systems, including Adaptive RAG, which uses reinforcement learning to dynamically control retrieval based on query complexity—achieving higher accuracy with fewer retrievals. Complementing this, my earlier work such as VulnGraph, SecureFixAgent, MalCodeAI, and MLCPD focuses on AI for code security, graph-based reasoning, and cross-language software intelligence.
My master's thesis, AI That Detects, Validates, and Repairs, advances a unified autonomous vulnerability lifecycle framework integrating detection, execution-based validation, and iterative remediation using hybrid LLM + symbolic reasoning pipelines. The goal is to build reliable, interpretable, and deployable AI systems for software security.
Beyond research, I have experience as a freelance software developer, delivering end-to-end systems including a Python-based billing and inventory platform for industry use. I have also mentored students as an NVIDIA Jetson AI Project Coordinator, guiding hands-on projects in deep learning and edge AI deployment.
Overall, my work bridges research and engineering, turning rigorous ideas into AI systems that are powerful, trustworthy, and shippable.
My path has been shaped as much by personal discipline as by academic ambition. As a martial artist, I hold a black belt in karate and earned a gold medal representing India at the International Karate Championship, 2017. Years of practice instilled in me mental resilience, strategic focus, and composure under pressure—qualities that deeply influence my approach to challenges.
Throughout school, I gravitated toward leadership, receiving Student of the Year honors three years in a row (grades 8-10) and serving in roles such as Junior Sports Leader and Class Representative. These early experiences taught me the importance of responsibility, teamwork, and effective communication.
My time with AIESEC Baroda's Outgoing Social Sector team exposed me to global development programs and youth-led social initiatives. Facilitating cross-cultural exchanges broadened my understanding of international collaboration, community impact, and the value of working toward a greater good.
Each chapter—from the dojo to the classroom to community leadership—has added a layer to who I am. I believe technical work is most impactful when rooted in empathy, discipline, and a drive to contribute meaningfully. These values guide the research I do and the way I engage with the world.