• Worked with clinicians and conducted research on leveraging ophthalmologists' gaze patterns to inform Vision Transformer (ViT) models for improved glaucoma detection accuracy.• Built a foundation model for Optical Coherence Tomography (OCT) images using Mask Autoencoder. Used the model with attention-based LSTM to develop a guiding system for novice ophthalmologists, enhancing eye disease diagnosis capabilities.• Building Multimodal LLM for referring-and-grounding biomedical images.

• Working on enhancing grounding capabilities for ophthalmic images as part of a collaborative project between Columbia University and Yale University.

• Assist in teaching graduate course COMS 4771 - Machine Learning.

• Researched and adapted image processing algorithms (image features extraction and blending) to improve image stitching quality in Surround View Monitoring System (SVM) project.• Researched and developed a solution for Gaze Estimation. Collaborated with other members to fuse it with the Forward Collision Warning system in C++ for DGX Xavier. The project was showcased at CES 2022, Las Vegas.• Innovated a new technique and formulated a corresponding algorithm in Python for the world-first Automatic Mirror Adjustment solution demonstrated in two VinFast Electric Vehicles at CES 2023, Las Vegas.

• Developed a web application for personal book reading, relevant information tracking, hiring and lending among users using MongoDB, Express, React and Node.• Applied CI/CD principles in deploying company's legacy web application and automating unit testing using Docker and Kubernetes, improving overall development and deployment cycle proficiency by 25%.

• Developed a personal knowledge management system in C# where each data type (text, image, audio) is managed by a dedicated microservice built on Azure serverless technologies (Azure Functions and Cognitive Services).• Configured and deployed these separated microservices as a web application using Azure DevOps.