RapidFire AI - ML Experimentation Platform (2024-Present)
As a founding engineer, I built the core ML experimentation UI platform from open-source MLflow, enabling researchers to visualize and manage model training across hyperparameters, architectures, and datasets. I architected AWS EKS deployment infrastructure that manages hundreds of concurrent clusters running multiple simultaneous training jobs. The platform includes real-time monitoring, experiment tracking, and an "Interactive Control" feature that allows users to clone and modify model layers during training, speeding up experiment accuracy up to 5x.
Project Quench - Healthcare RAG Application (2023-2024)
I wrote the entire backend and frontend MVP for a platform that transforms thousands of patient medical records into a semantic search engine with document references. The system allows physicians to ask questions and organize cases with low hallucination rates. Built using Python + FastAPI backend, React frontend, and automated OCR/embedding pipeline with AWS Textract and Llama 2. Delivered working version in four weeks, leading to first dozen customers including expert physicians at Harvard and UCSF.
BookShelf iOS App
Personal project that grew to 5k users and ranked #138 in the App Store Books category. Built the backend infrastructure for collections sharing feature and internal analytics. The app helps users track their reading progress, organize books into collections, and discover new books through AI-powered recommendations.
Samsara ELD Compliance Platform (2022-2023)
Maintained >99% SLA for the core ELD (Electronic Logging Device) feature used by 100,000+ drivers. During a team transition period, I took on a critical two-year-old project for updating GPS logging logic that would have caused compliance failures. Redesigned the system architecture and delivered the solution in under a month, preventing a major compliance crisis.
Expensify Money Transfer Platform (2020-2022)
Led development of money transfer features for Expensify's business-chat platform (think Venmo meets Expensify). Managed a team of 4 engineers and led the rollout to 100% of customers. Also led critical customer migration for 2000 of the biggest corporate customers using NetSuite integration.
I calculated and visualized popular clothing trends according to categorized Instagram posts from a Cornell dataset using d3.js and Python to process the clothing attributes. I worked alongside a WashU classmate, Vihar Desu.
I visualized geoclustered opioid drug transactions in my home state of NJ with two other WashU students, Ben Choi and Jessie Korovin. Our work was recognized as the top-ranked project in our "Cloud Computing with Big Data Applications" class. We used Spark (running on Amazon EMR), Python, and matplotlib to perform a k-means clustering on the data.
Over the summer, I interned with a team of engineers at eluv.io, an SF-based video and blockchain startup, to demonstrate the superior performance of their video streaming platform using React.
This year, I also participated in the CalHacks hackathon with a friend from UC Berkeley, Frank Wang, on a ride sharing challenge. We visualized ride-sharing pickup and dropoff data in Chicago to get a mapping of how to optimize ride-sharing fleet software.
I created a real-estate software MVP with a WashU grad student Andrew Emory, that scanned in client contracts, then populated a personalized calendar with relevant events and sent out scheduled emails to the necessary contractors. We built it using Django, a free OCR API, Amazon s3 and Bootstrap.
I've won several hackathon competitions (both independent and on teams) in various categories, including 1st Place Eluvio Content Fabric Challenge @ CalHacks 2018, 1st Place Microsoft Azure @ Building Blocks 2018, 1st Place PubNub Civic Chatbot @ NYC Developer Week 2017. See my Devpost profile for more details.