Customer Personal Information Management Application - Developed RESTful API's using Java and Spring Boot to gather data from 15-20 datastreams and reducing maintance work completely using a config driven framework.Created automated testing framework with Cucumber and Junit to increase code coverage from 0 to 95%.Designed and made API health monitoring from scratch using Cloudwatch , DynamoDB and Splunk .Created automation scripts using Databricks and DynamoDB to supply customer… Show more Customer Personal Information Management Application - Developed RESTful API's using Java and Spring Boot to gather data from 15-20 datastreams and reducing maintance work completely using a config driven framework.Created automated testing framework with Cucumber and Junit to increase code coverage from 0 to 95%.Designed and made API health monitoring from scratch using Cloudwatch , DynamoDB and Splunk .Created automation scripts using Databricks and DynamoDB to supply customer personal information to teams efficiently to eliminate the need of a program back office saving the company 4 million/year.Test Account Data Management -Built and maintained data management platform to create test credit cards used across company. Automated account creation from a manual day long proccess to seconds. Set up scripts using Databricks to automatically condition users nightly.Automated process for conditionining credit cards to easily make comprimised accounts to be used in the fraud space for testing.Created CLI to handle batch processing of API calls managing credit card conditioning using Javascript and NPM , reducing conditioning time by 100%.iOS Development of Fraudulent Charge Features -Spearheaded the design and development of robust fraud and disputes screens for the enterprise ios application, utilizing Swift , UIKit and Combine saved the company 95 million dollars every year for unresolved fraud/dispute transactions.Led the development of a disputes tracking feature that allows 55 million users to display the status of their credit card disputes.Increased code coverage from 20% to 95% by creating automated UI tests using test harness2023 Elected Site Lead for Tech Development Program Show less

Developed features for an internal web application to centralize the string translation process at Capital One. Application made using Angular front end and Kotlin/AWS Lambda backend. Features help reduce string translation process from 4-6 weeks to a couple days.

Developed informative user interface for Capital One clickstream data using Python/Dash for front end and Postgres/RDS for the backend service layer. Center goal is to generate maximum value each visit for Capital One teams to use clickstream data available on the UI. Exposure to AWS and Agile Development.Participated in company hackathons and Capital One Coders (Volunteer mentor for middle school coding students)

Undergraduate research focused on the dynamics of stellar systemsAssist in developing machine learning analysis for gravitational forces using NEMO.

• Implemented and Created Contextual AI Assistant using Rasa API, Unix and Java with 95% confidence understanding member concerns through a chatbot. This was done using the Rasa NLU API to understand the intent of a given message, then using Rasa Core API to carefully curate a reply that would assist members. • ML Language Models compared and tested: Scikit-Learn, Keras, BERT and, spaCy.

• Tested and modeled a homomorphic cryptosystem using Python/Java and Unix, as measured by a project to display effectiveness of the cryptosystem by the computation limits of a raspberry pi. This was done by first understanding how the cryptosystem worked, then choosing a low-form factor device to display the efficiency of the algorithm. • Raspberry pi was capable of homomorphic private information retrieval as well as classifying acoustic signatures using a simple AI algorithm built with… Show more • Tested and modeled a homomorphic cryptosystem using Python/Java and Unix, as measured by a project to display effectiveness of the cryptosystem by the computation limits of a raspberry pi. This was done by first understanding how the cryptosystem worked, then choosing a low-form factor device to display the efficiency of the algorithm. • Raspberry pi was capable of homomorphic private information retrieval as well as classifying acoustic signatures using a simple AI algorithm built with Scikit-Learn. Show less