About Me
I am a first-year masters student at Georgia Tech, majoring in Computer Science with a specialization in Machine Learning. My current research focus is understanding how to create more efficient 3D reconstruction models and understanding how to use their abilities to improve various fields, such as medical imaging.
Previously, I worked for The Cigna Group as an AI Engineer Intern, where I utilized open-sourced models or enterprise-ready LLMs to create data ingestion and generations applications to empower various business operations in the company.
I was also an Undergraduate Researcher throughout my four years at my alma mater, Rensselaer Polytechnic Institute, where I engaged primarily in Computer Vision-oriented projects such as utilizing Neural Networks to identify if videos of mock surgical procedures were successful for unbiased proctoring. In my spare time, I enjoy drawing, animating, and reviewing movies.
Education
Masters of Science in Computer Science
Georgia Institute of Technology
Bachelor of Science in Computer Science
Rensselaer Polytechnic Institute
- Graduated Magnum Cum Laude with 3.85 GPA
- On Dean's Honors List for all semesters
- Recieved Rensselaer Leadership Scholarship, which is given to 10% of accepted students
- Brother & Eboard Member of Alpha Phi Omega (Community Service Fraternity)
Experience
AI Engineer Intern
The Cigna Group
- Worked on the Applied GenAI Team, building enterprise-wide tools to enhance data engineering and business analytics workflows
- Refactored a Python Text-to-SQL algorithm with an ORM library to enforce schema on multiple Neo4j knowledge graphs, enabling CI/CD for a core GenAI database.
- Developed UI over relational metadata in Neo4j and deployed to Elastic Kubernetes Service, optimizing performance by eliminating need for Cypher Querying Language, saving ~$2400 annually by providing read-only Role-Based Access Control
- Developed DAG in AWS MWAA to perform stress tests to evaluate ability to connect with services such as Redis and Oracle, expanding connectivity by 77% and substantially improving an environment used by 20+ engineers
- Developed Python LLM-powered SQL Analytics Algorithm to extract data on join conditions in PySpark and Teradata SQL files and ingest into Neo4j knowledge graph databases using Cypher and Kubernetes
- Implementing the capture of 2623 stored procedures, 10582 SQL statements, and 77378 join conditions, that will be used as metadata for enterprise-wide text-to-SQL algorithm to improve join condition generation ability
- Designed and implemented a React UI for a Knowledge Graph management tool, used by 10+ enterprise teams to manage and visualize their Neo4j graphs
- Collaborated with multiple teams across the company to plan and define use cases for a new LLM-powered data generation application
Data Engineer Intern
The Cigna Group
- Developed a PySpark proof of concept to ETL customer call data from 50+ AWS S3 buckets into Databricks Delta Tables, enforcing schema and improving data parsing for analytics pipelines
- Reduced ETL processing costs by 21% by migrating Teradata SQL stored procedures to PySpark/SparkSQL in Databricks as part of a company-wide cloud migration initiative
- Engineered a scalable backend REST API using Node.js, Express, and MongoDB for a drug pricing application, simplifying data retrieval for front-end developers
- Presented the product's backend architecture in a company-wide demonstration, earning positive feedback from judges and leadership
Projects
GitCub
Allows developers to compare commits from repositories and recieve AI-generated insights on changes, such as summaries of changes, potential bugs, security risks, performance improvements, and sorting changes into pre-set categories.
I served as the backend lead, where I managed a team of 5 interns through the design and implementation of our FASTAPI backend and AWS resources (Lambda, APIGateway, VPC, DynamoDB). I also connected our backend to our Next.js frontend and ensured smooth communication between the two.
Our project recieved positive feedback from developers on the detail of the insights and functionality of the application and is projected to save ~10,000 per 100 engineers in development costs.
MyCity
A patent-pending internal networking application designed to connect employees. The platform was built with a full-stack, cloud-native architecture, featuring a serverless back-end and a responsive web interface to facilitate seamless user interaction
I served as the back-end lead and database administrator. I spearheaded the architecture and management of a DynamoDB instance and led a team of 5+ interns in developing a REST API using TypeScript in AWS Lambda. Additionally, I managed the integration of our API endpoints with over 10 front-end web pages built with Next.js, enabling full end-to-end functionality
The serverless back-end I led the development of was highly cost-efficient, with estimated yearly operational costs under $1000. The project was presented in a company-wide showcase, where the back-end implementation received positive feedback from judges for its design and functionality
Audio HCI System
A human-machine communication interface designed for research projects. The system leverages cloud-based AI to automate interaction by interpreting a user's verbal cues and generating a synthesized audio response, enabling real-time, voice-driven experiences
I created the interface by integrating services from Google Cloud. My responsibilities included architecting the system to process natural language, identify verbal cues, and trigger the appropriate synthesized audio output
This interface became a key component for other research initiatives, including its successful implementation in an augmented reality (AR) game designed to teach users Mandarin. Its functionality provided a practical and effective way to build interactive, educational applications
Research
Graduate Researcher
Georgia Tech Viola Lab
- Utilizing deep learning and 3D reconstruction to create more efficient compressed representations of 3D meshes
Undergraduate Researcher
RPI Radke Lab
- Organized and labeled training and testing datasets of airport security footage to detect potential anomalies
- Developed supervised machine learning models, RAG pipelines, and traditional computer vision approaches to detect anomalies, leveraging time, crowdedness levels, and behavioral patterns
Undergraduate Researcher
RPI CeMSIM Lab
- Implemented a classifier 1D convolutional neural network using OpenCV and PyTorch with 88% test accuracy rate to identify whether recorded surgical procedures were successful for student assessments
- Expected to be used in research project and potential paper as part of joint study between RPI and University of Buffalo
Undergraduate Researcher
RPI CISL Lab
- Created a human-machine communication interface using Google Cloud systems to efficiently automate interaction by reading verbal cueues and generating an audio response
- Expected to be used in multiple lab projects, including augmented reality game to learn world languages
Undergraduate Research Assistant
RPI CogWorks Lab
- Organize metadata generated from volunteers playing Tetris to gain insights into human’s cognitive processes, specifically how they react to the pressures of situations with increasing difficulty
- Helped design and test drive challenge levels of increasing complexity for an experimental game which tracks the correlation between individual cognition and group behavior
- Assisted in a pilot implementation of this game using 2 groups of 4 volunteers to create a database of eye movements and player actions over multiple sessions