File Systems and Storage Lab (FSL), Stony Brook University
- Analyze Shamir's Secret Sharing and related schemes to evaluate efficiency, robustness, and integrity
guarantees for long-term storage systems.
- Use Rust libraries and formal cryptographic arguments to bound adversarial advantage to negligible levels.
DIENS Lab, Stony Brook University
- Researched zeroth- and first-order optimization methods for machine learning in non-convex landscapes.
- Implemented PyTorch frameworks for SGD and Adam variants, improving convergence on high-dimensional
datasets.
- Developed mathematical models combining convex optimization theory with neural network convergence
analysis.
Arlington Computational Linguistics Lab, UT Arlington
- Curated dataset of 250K+ feline images; trained age-classification models using TensorFlow and PyTorch.
- Built an end-to-end ML pipeline for 100+ GB of video, extracting 20K+ vocalizations using YOLOv8 and
PANNs.
- Designed and evaluated acoustic feature sets for developmental vocalization modeling.
NASA Goddard Institute for Space Studies (GISS)
- Presented climate change analysis at AGU23 using GIS data [abstract].
- Processed 10+ GB of environmental data with Python, Pandas, and Google Earth Engine.
- Applied clustering, decision trees, and random forests to predict harmful algal bloom risk.
