• Led the migration of Oracle’s receipt recognition product from traditional ML models to a GenAI-powered, multi-tenant architecture, improving scalability and delivering 15–20% accuracy gains in core expense fields. This directly advanced Oracle’s Touchless Expenses initiative, enabling customers to create expenses automatically by simply uploading or photographing receipts.

• Contributed to the mass architectural shift toward the new BOSS architecture, refactoring legacy codebases to support Oracle’s latest expense automation platform.

• Designed and delivered agentic flows for unstructured document uploads, automating parsing and structured extraction across PDFs, images, emails, and other formats.

• Developed cloud-native microservices and REST APIs (Java, Helidon, Kubernetes) integrated with Oracle ERP, ensuring scalability, reliability, and enterprise adoption.

• Collaborated with cross-functional teams (engineering, product, QA) to deliver features end-to-end, from backend services and GenAI pipelines to model evaluation and customer-facing workflows.

• Completed the following relevant coursework: Artificial Intelligence, Applied Deep Learning, Natural Language Processing, Computer Vision, Cloud Computing, Advanced Software Engineering, Advanced Database Systems, Computer Networks, Analysis of Algorithms, 3D UI and Virtual Reality

• Devised and managed AWS cloud architecture for AI-based skin evaluation system at MedySapiens, a biotech startup in Series-A

specializing in cognitive computing for healthcare

• Expedited labeling medical image dataset by integrating AWS Sagemaker into workflow

• Improved classification accuracy of skin damage in new image dataset by 5%

Columbia Computer Graphics and User Interfaces Lab (CGUI)

• Developed a VR platform with Unity that allows multiple users to simultaneously interact with a map embedded with geotagged data (social media, real-time traffic, etc)

• Integrated Instagram and Yelp data to the platform and created map navigation functionalities

Columbia Computational Design Lab

• Built a Proof of Concept web application on which scientists can publish their work in the form of Twitter threads

Columbia Computer Vision Laboratory (CAVE)

• Designed and deployed database of news articles to AWS cloud for the Cross Cultural Analysis Project, which aims to distinguish cultural differences between two countries presented in video news

• Implemented automation of keyword extraction and its analysis from video data

• Optimized search queries for video data collection by providing blacklist and whitelist words

• Enhanced test frameworks for computer vision applications on Snapdragon 855 chip for Samsung Galaxy S10

• Generated Python test scripts for DENSO robot arm to physically test the Snapdragon X50 5G modem

• Initiated and composed internal Python programming guide for the DENSO robot arm