PhD in Information Systems (AI/ML) · UMBC · Baltimore, MD
Zahid
Hassan
Tushar
Building foundation models that make Earth observation more efficient.
scroll ↓
PhD in Information Systems (AI/ML) · UMBC · Baltimore, MD
Building foundation models that make Earth observation more efficient.
I completed my PhD in Information Systems (AI/ML) at the University of Maryland, Baltimore County, where I worked at the intersection of foundation models, geospatial intelligence, and trustworthy AI.
My research develops parameter-efficient, physics-informed architectures for high-dimensional satellite data — with applications in atmospheric cloud and aerosol property retrieval from NASA's PACE mission. I am actively seeking Postdoctoral and Assistant Professor positions starting 2026.
Developing compact, parameter-efficient foundation models for hyperspectral satellite data using spectral grouping and representation compression. Enables scalable Earth observation with physical consistency.
Designed attention-based and generative architectures (Transformers, UNets, Diffusion Models) to solve atmospheric inverse problems, studying robustness under multi-angle geometry and 3D radiative effects.
Surveying threats, defenses, and open challenges in the security landscape of large language models applied to biomedical domains.
Investigated robustness of federated learning under adversarial attacks for automatic speech recognition, studying privacy risks in distributed model training.
Built a self-supervised framework for skin cancer classification and designed explainable AI methods to interpret CNN representations and decision patterns.
Contributor to NASA-funded open-source research software for operational satellite analytics, with packages released through the NASA ACCESS program.
PyTorch · GANs · Diffusion Models · Transformers · CNNs · UNets · LLMs · Self-Supervised Learning · Foundation Models · Explainable AI
Regression · Classification · Regularization · Model Selection · Bias-Variance Tradeoff · Evaluation Metrics
High-Dimensional Data Modeling · Multimodal Data Fusion · Representation Learning · HPC Clusters · Multi-GPU Training
GitHub · Reproducible Pipelines · NASA ACCESS Program · Open-Source Research Software
Generative AI · Machine Learning · Data Science · Deep Learning · Data Visualization · Python Programming · AI for Scientific Discovery · Trustworthy AI · Federated Learning
Signals & Systems · Electronic Engineering · Electrical Machines · Digital Signal Processing Lab · Communication Systems Lab · Digital Electronics Lab
Mentored undergraduate and K–12 students in AI research projects at UMBC. Supervised laboratory instruction and undergraduate research at MIST.
Extract information from PDF files, store as JSON/YAML, and chat about the content using a free local LLM — no API key required.
I am actively seeking Postdoctoral and Assistant Professor positions starting 2026. I welcome inquiries about research collaborations, speaking invitations, and faculty opportunities.
Current Position
PhD
University of Maryland Baltimore County
Department of Information Systems
Advisor
Prof. Sanjay Purushotham
Graduation
Fall 2026