Varun Sonawane

Featured Projects

Solutions I've built to solve real-world problems

IRIS

Honorable Mention Claude Hackathon

Intelligent Retail & Ingredient Scanner

An AI-powered grocery assistant that bridges personal health data with real-time food decisions. Users upload lab reports and scan product labels to receive personalized verdicts — Safe, Caution, or Avoid — powered by Claude's vision and reasoning.

Key Achievements

Built at the Claude Builder Club Hackathon by Indiana University students
Lab report analysis extracting health markers like cholesterol and allergies
Real-time product scanning via camera with instant Safe/Caution/Avoid verdicts
Voice feedback via ElevenLabs for hands-free shopping and smart alternative suggestions

Technologies

Claude (Anthropic)

ElevenLabs

React

Vite

Node.js

Express

Tailwind CSS

Python

JavaScript

DataLens

Google Gemini 3 Hackathon Project

Multimodal AI Data Storytelling Platform

Code Live Demo

An AI-powered platform that transforms raw CSV/Excel data into rich, multimodal stories using Google Gemini. Features a real-time voice agent for hands-free data exploration, RAG-powered conversations, and parallel image generation woven into narratives.

Key Achievements

Live deployment on Google Cloud Run with full async FastAPI backend
Triple-perspective storytelling (ELI5, Architecture, Analyst) streamed in real-time
Gemini Live Voice Agent via bidirectional WebSocket for conversational data exploration
RAG pipeline with Vertex AI Vector Search (text-embedding-004, 768 dims)

Technologies

Google Gemini

FastAPI

React

TypeScript

Vertex AI

Cloud Run

GCS

Firestore

WebSockets

SSE

VoiceLegal AI

GCP Hackathon Project

AI-Powered Legal Document Assistant

Code Live Demo

A Chrome extension that analyzes legal documents instantly using Google Gemini 2.5 and allows users to ask questions via voice interaction powered by ElevenLabs.

Key Achievements

Built for AI Partner Catalyst Hackathon 2025
Integrated Google Gemini 2.5 Flash for analysis
Real-time voice interaction with ElevenLabs Conversational AI
Seamless browser integration via Chrome Extension

Technologies

Google Gemini

ElevenLabs

React

FastAPI

Chrome Extension

Docker

GCP

EngageIU

Luddy Hackathon Winner

Campus Event Attendance Leaderboard

Code Live Demo

A real-time campus event engagement platform built for a 24-hour hackathon. Students earn points by entering event check-in codes, with live leaderboard updates via Server-Sent Events and a JWT-protected admin dashboard for full event management.

Key Achievements

Real-time leaderboard via Server-Sent Events with live connection status indicator
Achievement badges (Top, Rising, Dedicated, Consistent) with week-over-week rank trends
JWT-protected admin dashboard with full event CRUD and QR code generation
Cryptographically secure check-in codes with timestamped attendance logging and CSV export

Technologies

FastAPI

PostgreSQL

Docker

Python

SQLAlchemy

Server-Sent Events

JWT

HTML

CSS

JavaScript

DocWeaver

Multi-Agent Clinical Intelligence Platform

Code Demo

A multi-agent AI orchestration platform that transforms clinical workflows through intelligent document processing. Uses 20+ specialized AI agents working in parallel to analyze medical records, generate SOAP notes, and automate care coordination.

Key Achievements

20+ specialized AI agents for each distinct clinical task
Multi-source data fusion, processes 5+ documents in parallel with temporal analysis
Generates complete SOAP notes automatically from brief clinical notes
Automated referral letters, follow-ups, and patient education materials

Technologies

Google Gemini

FastAPI

Next.js

TypeScript

Python

Streamlit

Async Processing

IdeaGenie

2nd Prize Winner

LLM-Powered Innovation Ranking Engine

Code Demo

Architected an end-to-end idea evaluation engine using Flask, Docker, and Qwen LLM via Ollama, implementing the ReAct (Reasoning + Action) framework to simulate iterative reasoning steps for multi-criteria scoring.

Key Achievements

2nd Prize – Hackathon Winner out of 50+ teams
Handled 100+ idea inputs with sub-180ms latency
Hybrid pipeline combining vector similarity with LLM evaluations
Production-ready REST API with input validation and concurrency

Technologies

Flask

Docker

Qwen LLM

Ollama

ReAct Framework

Python

REST API

React

AWS Big Data Pipeline

End-to-End Health Data Processing System

Code

A scalable big data pipeline handling 100,000+ health records using distributed PySpark on AWS EC2. Features automated ML training via SageMaker Autopilot and serverless event-driven processing.

Key Achievements

Processed 100k+ records with distributed PySpark on EC2
Automated ML model training using AWS SageMaker Autopilot
Serverless data pipeline with AWS Lambda & S3 triggers
Discovered 72% population risk via Power BI analytics

Technologies

AWS SageMaker

PySpark

AWS Lambda

EC2

Power BI

Python

ShieldScraper

AWS-Based Automated Data Pipeline

Code

Developed a secure, fully automated web scraping pipeline leveraging Scrapy for data extraction, containerized with Docker, and deployed on AWS ECS Fargate for scalability and reliability.

Key Achievements

Fully automated web scraping pipeline
Real-time monitoring with CloudWatch and SNS
Interactive analytics dashboards with QuickSight
Daily automated workflows with EventBridge

Technologies

Scrapy

Docker

AWS ECS Fargate

AWS Glue

Lambda

DynamoDB

Athena

QuickSight

CloudWatch

ETL Pipeline for Weather Data

Airflow and Docker Implementation

Code

Developed an automated ETL pipeline to extract real-time weather data from the OpenWeather API, transform it using Python, and load it into a PostgreSQL database, processing data for over 10+ locations daily.

Key Achievements

Automated ETL pipeline for 10+ locations daily
Containerized deployment with Docker
Modular DAGs for workflow automation
Seamless analysis and visualization capabilities

Technologies

Apache Airflow

Docker

PostgreSQL

Python

OpenWeather API

ETL

Data Visualization Dashboard

Interactive Analytics Platform

Code Live

Built a comprehensive data visualization platform with interactive charts and real-time analytics. Features dynamic dashboards, advanced filtering, and multi-source data integration for comprehensive business intelligence.

Key Achievements

Interactive dashboards with real-time data updates
Multi-source data integration and processing
Advanced filtering and drill-down capabilities
Responsive design with mobile optimization

Technologies

D3.js

JavaScript

Python

Flask

PostgreSQL

Data Visualization

More Projects

Term Deposit Subscription Prediction

Machine learning model to predict customer term deposit subscription using classification algorithms and data analysis.

Python

Scikit-learn

Pandas

Machine Learning

Classification

Stock Market Prediction System

Advanced stock market prediction system using time series analysis and machine learning for financial forecasting.

Python

TensorFlow

Time Series

Financial Analysis

Deep Learning

Rock Paper Scissors Game

Interactive rock paper scissors game with modern UI design and game logic implementation for entertainment and learning.

JavaScript

HTML5

CSS3

Game Development

DOM Manipulation

Explore More Projects

Check out my GitHub profile for more projects, contributions, and open-source work.

View GitHub Profile

Experience

My professional journey and key achievements

Associate Instructor, Data Science

Indiana University Bloomington, Luddy School of Informatics

Bloomington, Indiana, USA

Jan 2025 – May 2026

Academic

Key Achievements

Supported instruction for two graduate and undergraduate data science courses with a combined enrollment of 150+ students.
Facilitated weekly office hours and graded assignments for DSCI-D595 (Data Science On-Ramp), a graduate-level course covering Scala, Apache Spark, Tableau, NLP, and web scraping for ~30 students.

Technologies Used

Scala

Apache Spark

Tableau

NLP

SQL

Data Engineer Intern

The Commons XR

San Diego, USA

May 2025 - Aug 2025

Internship

Key Achievements

Achieved real-time, cross-system data syncing, as measured by a 100% reduction in manual handoffs and faster reporting cycles, by building data pipelines from PostgreSQL to both PostgreSQL and BigQuery, and implementing CDC using Google Cloud Datastream.
Improved query maintainability and execution reliability, as measured by a 40% drop in SQL runtime errors and smoother dev handoffs, by refactoring legacy SQL into modular SQLx files and integrating them into Python-based BigQuery pipelines.

Technologies Used

PostgreSQL

BigQuery

Google Cloud Datastream

SQLx

Python

Plotly

Software Engineering Intern

CodeClause

India

July 2023 - Jan 2024

Internship

Key Achievements

Increased backend system reliability by reducing recurring service failures across 5+ microservices by implementing unit tests and integrating CI/CD pipelines.
Delivered a high-impact user-facing feature by building and integrating a React-based frontend with REST APIs, improving usability.

Technologies Used

React

REST APIs

CI/CD

Infrastructure as Code

Microservices

Unit Testing

Technical Lead

Unstop Igniters

India

Mar 2022 - Jul 2023

Leadership

Key Achievements

Implemented structured project management for 4 hackathons and 8+ coding competitions, enhancing team work while showing attention to detail in requirements gathering across technical challenges.
Coordinated 10+ member teams for 5 industry speaker sessions and 7 technical workshops, defining clear project scope and success metrics while demonstrating communication, organization, and problem-solving abilities.

Technologies Used

Project Management

Team Leadership

System Architecture

API Design

RESTful Services

Containerization

Data Science Intern

CodeClause

Pune, Maharashtra, India

Feb 2022 - Mar 2022

Internship

Key Achievements

Led the project 'Predicting Term Deposit Subscription,' showcasing proficiency in data science through extensive research and advanced analytical techniques.
Employed web scraping for comprehensive data acquisition and implemented advanced data cleaning techniques to ensure high-quality datasets.

Technologies Used

Data Analytics

Statistics

Python

Machine Learning

Data Science

Web Scraping

Ethical Hacking & Cyber Security Intern

Supraja Technologies

India

Aug 2020 - Sep 2020

Internship

Key Achievements

Specialized in Ethical Hacking and Cyber Security, gaining expertise in identifying system vulnerabilities through intrusion evasion, firewall management, and honeypot analysis.
Honed skills in proposing effective mitigation strategies through comprehensive ethical hacking exercises and security assessments.

Technologies Used

Kali Linux

Ethical Hacking

Cybersecurity

Nmap

Wireshark

Tenable Nessus

Currently seeking full-time opportunities

Technical Skills

Technologies and tools I use to build scalable solutions

Cloud & Infrastructure

AWS

Azure

Docker

Kubernetes

Terraform

CI/CD

Data Engineering

SQL

PostgreSQL

MySQL

MongoDB

Redis

Data Modeling

Programming Languages

Python

JavaScript

Java

C++

TypeScript

ML & LLMs

Scikit-learn

TensorFlow

LangChain

Prompt Engineering

ReAct

Big Data & ETL

Apache Airflow

PySpark

Spark

Hadoop

Batch Processing

Visualization & Tools

Power BI

Tableau

Git

Linux

VS Code

System Design

About Me

My journey from curiosity to scalable engineering

I started with a B.E. in Information Technology from Savitribai Phule Pune University, graduating with honors and a specialization in Data Science. Now I’m finishing my Master’s in Computer Science at Indiana University Bloomington (GPA 3.86/4.0), with coursework in Cloud Computing, Advanced Databases, and Applied Machine Learning.

My work sits at the intersection of data engineering and cloud architecture. I build pipelines that move millions of records, warehouses on Medallion Architecture, and AI systems that actually ship. From my first SQL query to deploying RAG pipelines on Vertex AI, the through line has been the same: turn a hard problem into something that scales.

Outside coursework, I’ve competed in and won hackathons, built full-stack data products with real teams, and picked up AWS and Azure certifications along the way. I’m at my best when the problem is messy and the solution has to be clean.