Projects
Git-aware Env Diff Tool in Go
Developed a command-line tool in Go to compare .env files across Git branches, helping developers catch configuration drifts early. Implements Git plumbing commands and dotenv parsing.
MedQuery: Text-to-SQL System for MIMIC-III
Built a Retrieval-Augmented Generation (RAG) system to convert clinical questions into SQL queries on the MIMIC-III dataset using FAISS, ChromaDB, and Sentence Transformers. Features Streamlit UI with ICD-9 code lookups and query history.
Observability in Motion: OpenTelemetry + GCP
Engineered observability for real-time BART transit data pipelines using OpenTelemetry, Pub/Sub, Dataflow, and BigQuery. Enabled end-to-end tracing, metric collection, and error diagnostics on GCP.
Real-time Energy Data Lake on GCP
Built a real-time Energy Data Lake using Google Cloud Platform by integrating GridStatus API data across 9 ISO grid systems. Used Dataproc, BigQuery, Pub/Sub, and Vertex AI to enable analytics and forecasting.
Last Mile Connectivity in DMV
Addressing last mile connectivity for public transport using - WMATA (Washington Metropolitan Area Transit Authority) Ridership Data, Capital Bikeshare Trip History Data, US Census Bureau Data and Transit App
EKS Open Telemetry
Automated deployment, monitoring, and observability for a 67-service application on AWS EKS, optimizing resource usage and system reliability.
Swiftly - ECommerce Web App on AWS
AWS-based robust e-commerce platform capable of managing high traffic loads with minimal downtime, safeguarding sensitive data, and optimizing global content delivery.
Adobe Analytics Challenge 2024
Led a team of 3 in the 2024 Adobe Analytics Challenge, analyzing customer journeys and vehicle purchasing behavior for General Motors using advanced statistical modeling and marketing analytics to optimize conversion rates.
Consulting - Community Forklift
Consulted Community Forklift on strategic recommendations to enhance operational efficiency, amplify community engagement, identify key performance indicators for their sales and optimize revenue generation via a data-driven approach.
NBA Expansion Proposal - Smith Analytics Consortium Datathon 2024
Participated in the 5th Annual Datathon hosted by the Smith Analytics Consortium in collaboration with Deloitte, as part of a 7-member interdisciplinary team. The competition involved a 15-day data-driven challenge to propose a new city for an NBA team, including selecting a sponsor, mascot, and international partnership.
GameFlix
GameFlix - Web app to showcase free-to-play cross platform games thegameflix.vercel.app/
MahaMetroLink
MahaMetroLink is a proposed system for Pune Metro as a one stop solution for customer functionalities and service desk operations developed during Salesforce Hackathon 2021 at Persistent Systems
InSight
Android Application for low vision community using Computer Vision and Text-to-Speech
DeepSpamReview
DeepSpamReview: Detection of Fake Reviews on Online Review Platforms using Deep Learning Architectures. Summer Internship project at CoreView Systems
Multilabel Toxic-Comment Classification
Multilabel Classification of Toxic Comments using BERT - Tensorflow HUB
ABU Robocon 2018
MATLAB Implemented Modules during ABU Robocon 2018