Brawner 249
Dr. Sarina Adeli is an applied AI researcher and Assistant Professor of Business Analytics and Information Systems at Menlo College, with expertise in generative AI, large language models (LLMs), multimodal learning, and geospatial machine learning. Her research examines the architectural, ethical, and operational dimensions of deploying LLM-based systems in real-world settings, including education, workforce development, and intelligent geospatial infrastructures.
Her current work investigates prompt sensitivity, cultural alignment, representation gaps, and trust calibration in LLMs. She designs and evaluates strategies for prompt engineering, retrieval-augmented generation (RAG), and in-context learning to mitigate hallucinations and model instability in open-domain and low-resource tasks. She also explores adversarial training and contrastive pretraining to improve robustness and controllability of transformer-based architectures.
Dr. Adeli previously worked at Apple. Prior to that, she collaborated with NASA JPL and the U.S. Geological Survey on national-scale satellite image classification using U-Net, LSTM, and ensemble CNN models. Her publications received over 1000 citations within two years of her graduation. Her work also includes multimodal document analysis and vision-language models for intelligent decision support.
She earned her Ph.D. in Data Engineering from the State University of New York, where her dissertation focused on scalable land cover mapping via multi-sensor fusion using SAR-optical integration and deep learning. During her M.Eng. program, she interned at the European Space Agency (ESA) funded project in Switzerland, contributing to SAR-based disaster response systems and synthetic data generation for model pretraining.
At Menlo College, she teaches courses on machine learning systems design, Python programming, and the foundations of GenAI. She is building a research pipeline to co-design multilingual, culturally responsive LLM agents for AI-mediated career advising and automated reasoning in STEM education. Her broader academic agenda aims to address frontier challenges in generative AI, such as output opacity, fairness in prompt outcomes, and self-consistency, while preparing students to work with and critique foundation models at scale.
