I am a Research Engineer building the future of artificial intelligence. My passion lies in transforming complex data into intelligent systems that make a real-world impact.
The Work
Currently, I work at Amazon Web Services in Santa Clara, where I focus on Agentic AI. I am building advanced Agent Simulation frameworks designed to test and validate AI agent behaviors in complex, dynamic scenarios. My work involves developing LLM-powered systems that can redact sensitive information in real-time while maintaining low latency, as well as optimizing machine learning models for audio transcript analysis.
The Background
Before joining Amazon, I was a Graduate Researcher at the Mark and Mary Stevens Institute of Neuroimaging at USC. There, I researched Tau PET imaging for early-onset Alzheimer's disease detection. I developed a novel Partial Volume Correction (PVC) algorithm that enhanced disease detection using patient MRI and PET images, leveraging Apache Spark for efficient distributed computing.
Prior to that, at the ITEMS Institute, I developed U-Net based models for axon and myelin segmentation from microscopic nerve data, optimizing CNN architectures to improve segmentation performance.
Selected Projects
Achieved rank 2/1946 in a Kaggle competition by predicting protein-protein interaction bindings using encoded molecular structures and 1D-CNN models.
Developed a GAN-based model for blood vessel segmentation from fundus images, achieving a 38% parameter reduction and improved quantitative metrics.
Implemented a Conditional GAN model for efficient haze removal from ADAS images, significantly improving real-time visual detection.
Used Transformer-based models with NLP techniques for 3D keypoint sequence classification to identify fingerspelled phrases.