Soham Jadhav

P.E.S. Modern College of Engineering, Pune

AI & Cloud Intern

• Completed a 4-week industry internship focused on AI and cloud deployment workflows
• Deployed machine learning models and cloud services using IBM Cloud and Microsoft Azure
• Collaborated in a team-based environment following structured project workflows and code review practices

Online graphics.h Compiler and VS Code Extension

• Built a browser-based compiler that runs legacy graphics.h C programs entirely client-side, removing the need for local Turbo C or DOSBox setup
• Developed a companion VS Code extension, now with 5,000+ installations, that lets students run graphics.h programs locally with a single click
• Grew the platform to 30,000+ active users, primarily engineering students working through legacy C graphics coursework

NotifySPPU

• Built an automated scraping pipeline that polls SPPU's result portal every 10 minutes, tracking 5,000+ student records across 100+ courses
• Designed a PostgreSQL schema to diff incoming results against stored records so only genuinely new results trigger a notification
• Shipped real-time browser push notifications via OneSignal, serving 1,000+ users the moment results are declared
• Scheduled and deployed the scraper with GitHub Actions for reliable, cost-free continuous polling

SPPUCodes — Academic Code & PYQ Platform

• Built and maintained a full-stack platform for sharing academic code and previous-year question papers, now serving 2,000+ active students
• Designed the backend architecture end-to-end: REST APIs, PostgreSQL schema, and Cloudinary-backed file storage for code and document submissions
• Implemented authentication, search, filtering, and content management features, growing the catalog to 1,200+ entries
• Used Google Analytics to inform SEO and UX improvements across 2,000+ user sessions

Driver Monitoring System

• Trained a YOLOv11 object detection model on a custom dataset of 13,000+ images to identify unsafe driver behavior in real time
• Implemented multi-class detection covering seatbelt usage, smoking, drinking, and driver distraction
• Converted and optimized the model with ONNX Runtime, enabling low-latency inference directly on client hardware without a server round-trip

News Sense AI

• Built a news summarization microservice achieving 70% extractive accuracy, processing 1,000+ articles
• Integrated Hugging Face summarization models behind a lightweight Flask API
• Slimmed the Docker image from 9GB to 2GB and deployed it on Azure Container Apps with autoscaling for variable traffic loads

Languages: Python, C, C++, JavaScript, TypeScript

Frontend: React, HTML, CSS, Tailwind CSS

Backend: Flask, REST APIs, PostgreSQL, MySQL

Cloud & DevOps: Microsoft Azure (VMs, App Services, Blob Storage, Container Apps), Docker, Git, GitHub, CI/CD, GitHub Actions

Tools: Cron Jobs, Google Analytics, SEO