Developed a data pipeline that feeds cross-service signals into an AI model and implemented a daily evaluation job to measure treatment vs. control performance of model iterations across prompts.
Owned the end-to-end development for a feature addressing the annual 12% office room booking decline rate (~17M occurrences) across all Outlook users. Resolved bugs, maintained 99.9% availability and reliability, and facilitated knowledge transfer across engineering teams.
Engineered an intelligent booking pipeline to suggest optimal locations to organizers and participants in multiple locations. Implemented heuristics that takes room capacity, vicinity and familiarity into consideration to reduce scheduling friction, resource usage and booking errors.
Worked as a Full-Stack engineer on a feature that automatically frees up unused rooms when attendees do not show up, addressing a global 30-40% no-show rate and increasing office room utilization by 20-30%.
Implemented comprehensive telemetry across multiple features, including exception tracking, availability and reliability metrics, business logic data points, and structured logging.
Software EngineerThoughtWireSep 2021 – Jul 2022
Migrated ThoughtWire’s infrastructure to reduce the amount of Hazelcast grids to a singular instance, facilitating containerization and improving cloud readiness of the platform.
Designed a feature enabling ThoughtWire’s platform to handle SIP requests, parsing and transforming them into meaningful data for a SIP-processing component.
Software EngineerMunicipal Property Assessment Corp.Aug 2019 – Feb 2021
Lead the development of a web application that facilitates MPAC’s on/offboarding processes using Angular, Spring Boot and AWS, automating the process completely.
Optimized a lookup that identifies the different types of properties that MPAC assesses by migrating it to React and AWS, decreasing the cost to run the application.
Developed an application with a microservices approach, Spring Boot and React that pulls metrics from different avenues of MPAC’s systems for our data analysts.
Real-Time Facial Analysis
Developed a computer vision application using OpenCV, TensorFlow, and DeepFace for real-time face detection, emotion recognition, and similarity comparison.
UniHome.Me
A Node.js web application built in 24 hours at YHacks 2016 using Express, Bootstrap, and Esri Map API to help students find housing by institution.
