Focused on researching novel computer vision techniques and deep learning models for multimodal medical image registration, instance segmentation, and image synthesis.Involved in designing a cascaded architecture using Inception modules and a variant of TransMorph for deformable image registration in brain tumor patients.Engaging with multimodal datasets including T1, T1-CE, T2, and FLAIR images, optimizing information extraction and displacement field generation.

Model Development: Architected and implemented a universal image classification model using PyTorch and TensorFlow, handling six distinct data types (8-bit to 16-bit imagery), increasing processing efficiency by 70%.YOLOv8 Classifier: Engineered a state-of-the-art YOLOv8 classifier to enhance 16-bit grayscale image detection, resulting in a 65% improvement in object detection accuracy under diverse environmental conditions.Multi-Head AlexNet: Designed and developed a multi-head AlexNet model to perform concurrent classification tasks on custom datasets, improving task efficiency by 50%.Data Processing Automation: Automated preprocessing and normalization of large-scale datasets, including multi-frame TIFF stacks and various bit-depth images, ensuring high-quality input for ML models.Optimization and Mixed Precision Training: Leveraged PyTorch's GradScaler for mixed precision training and optimized training pipelines with Adam optimizer and learning rate schedulers, reducing training time by 40%.End-to-End ML Pipeline: Built an end-to-end ML pipeline encompassing data loading, transformation, model training, and validation using Python, YAML configuration, and argparse, automating the entire workflow.Integration and Deployment: Integrated advanced ML models into production systems, enhancing real-time object detection capabilities. Contributed over 5,500 lines of maintainable code within a Git-controlled environment.Visualization and Reporting: Developed scripts for real-time visualization of training metrics using Matplotlib, and generated comprehensive model evaluation reports, enabling data-driven decision-making.

Details found in project section below :)

Explored Applications of Machine Learning for Wildfire Monitoring and Surveillance using UAVs. Pitched final presentations to NASA AMES scientists.This paper documents exploratory efforts using machine learning to determine efficient flight paths for UAVs and to detect wildfires using image classification. On path planning, three machine learning techniques—Genetic Algorithm, Simulated Annealing, and Dynamic Programming—were explored.

Engaged in a one-month residential program focusing on Semiconductor Materials and Device Engineering. Explored topics including semiconductor doping, pn-junctions, manufacturing processes, and transistor fabrication. Conducted research on FinFET technology and showcased findings through a poster presentation at the program's closing symposium.

Among the top 2% of interns selected nationwide for extensive training on Google's product lifecycle, including ML, UX/UI design, and blockchain applications.Led a team to develop "ReStyle," a clothing sustainability app, designing 24 UI screens and integrating user authentication and payment processing using Google's technologies.Presented the "ReStyle" app concept at a pitch night to a panel of tech leaders and entrepreneurs.