Hi, I'm Shifa
I'm a Master's candidate in Data Science at the University of Maryland. I've always been drawn to work that sits between perception and reasoning — systems that have to understand where things are and when they happened before they can say anything useful about them.
Lately that's meant Vision-LLMs and spatio-temporal retrieval, multi-agent systems that turn plain English into running infrastructure, and 3D perception stacks that survive contact with a real field of crops. I turn research ideas into systems that ship, and I'd love to talk to anyone working further along this path.
Graduate Assistant, University of Maryland
Teaching calculus and statistics to a hundred-odd undergrads, and building Decode — an AI notebook that catches mistakes in handwritten work as you write it.
M.S. Data Science, University of Maryland
College Park, MD.
Runner-Up, AI Spatial Intelligence Hackathon
Hosted by Ironsite and UMD Startup Shell.
Runner-Up, AI & Food Insecurity Case Competition
Hosted by the Smith School and the Capital Area Food Bank.
Machine Learning Engineer, Tata Elxsi
Taught tractors to tell crops from road — a LiDAR perception stack that had to run in real time, on a Jetson, in an actual field.
Machine Learning Intern, Tata Elxsi
A computer-vision tool for rail-signalling validation that took most of the manual testing off the table.
Winner, CodeForGood Hackathon
Hosted by JP Morgan Chase.
B.Tech Computer Science, GITAM University
Visakhapatnam, India.
VISTA
A persistent memory layer for frontier models that understands egocentric video — tracking objects, spatial orientation, and temporal context through allocentric heading and scene graphs. Outperformed Gemini 3 Flash and Molmo-2 on spatial-temporal reasoning.
Cloud Orchestrator
A multi-agent system that translates plain English into executable GCP infrastructure DAGs across 20+ services (Vertex AI, GKE, BigQuery), cutting cloud setup time 95% through automated tool orchestration.
