Reya Vir

I am an engineer, researcher, leader, and athlete who is passionate about applying my knowledge to real-world problems. I have a broad interest in the fields of software engineering, machine learning, and cloud computing.

My Research

Laboratory of Professor Iain Carmichael, UC Berkeley & UC San Francisco
2022 - Present
• Conducting research with Prof. Iain Carmichael to advance cancer diagnosis through the application of machine learning techniques and supervised learning models in computational pathology
• Conducted comprehensive benchmarking of state-of-the-art machine learning models for histopathology image recognition and diagnosis, evaluating their performance on various metrics
• Currently working on an image registration project to find the optimal mapping between H&E and multiplex immunofluorescence slides of the same tissue. Exploring various algorithms to identify keypoints and features to determine homography matrices

Laboratory of Professor Niloufar Salehi, UC Berkeley EECS & School of Information
2023 - Present
• Conducting ML and HCI research to enhance doctor-patient communication. Actively involved in the creation of a question sequence aimed at facilitating comprehension post-diagnosis, ensuring patients can better navigate their healthcare journey. Focused on utilizing LLMs and targeted prompts to empower patients in asking a curated series of questions, tailored for optimal understanding of their medical condition.

Supervised Program for Alignment Research
2023 - Present
• Conducting research under the guidance of Chan Bae as part of the Berkeley AI Safety Initiative.
• Training Decision Transformers on gameplay strategies for chess, using reinforcement learning with game outcomes as rewards.
• Analyzing the attention mechanisms, activations, and policy gradients to better understand how goal-driven behavior emerges in large neural networks through self-play.

Laboratory of Dr. Gregory Goldgof, Memorial Sloan Kettering Cancer Center
2023 - Present
• Managing Cloud at Cal’s research project with Memorial Sloan Kettering Cancer Center to scale high-performance cell imaging and classification of blood and bone marrow cell classification and assist pathologists.
• Improving the hematopathology web application by creating AWS Lambda functions for the “tiling” process to allow low-latency concurrent cell processing to accelerate cell identification analysis from minutes to seconds.
• Deploying the containerized application onto a serverless architecture and migrating parts of the backend for processing images and running AI classifiers to AWS Lambda for improved performance.

Cloud at Cal, President (2023)/ Research Project Lead (2022-2023)
2022 - Present
• Manage our Development, Research, Education, and Competitions branches: leading company development projects, research projects, and our Cloud Computing course
• Next Frame Prediction Project: Led research team in the development of a ConvLSTM-based next-frame prediction model using web-scraped dashcam footage for self-driving cars.
• Utilized LSTM architecture to handle sequential data for real-time predictions, updating with new input frames.
• Implemented ground truth analysis to identify abnormal vehicular behavior, enhancing model accuracy.
• Currently developing an early warning system for deviations from predicted road conditions, contributing to improved safety in autonomous driving.

Laboratory of Professor Silvia Figueira, Santa Clara University
May 2020 - Sep 2020
• Worked under Dr. Figueira, Director of Frugal Innovation Hub at SCU. Researched mobile phone battery usage patterns across various languages and video watching methods. Learned React Native and created an app to track mobile phone battery usage