What I’ve built.
Scientific archives, authoring tools for educators, fine-tuned models, and a handful of course projects. Each one has a write-up of the problem, what I chose, and what it ended up doing.
A searchable archive for AstroSat.
A scientific data platform that helps researchers find, extract, and analyse observations from ISRO’s AstroSat archive.
Scientific archive made accessible
Course authoring without a developer.
A self-service authoring platform that lets subject matter experts create interactive learning modules, without waiting on a developer.
Sentiment across eleven Indian languages.
Fine-tuning Llama 3.1 to recognise sentiment across Indian languages, with 29 experiments and a single GPU.
11th of 223 competition participants
Everything else
12 projects
From 500-Pages of PDFs to Multilingual Learning in 30 Minutes
Engineered an end-to-end Python pipeline using Google Gemini and AWS Bedrock to automate the transformation of 300+ course PDFs into multilingual, structured learning modules with >97% semantic accuracy.
Learn By Doing - Authoring Platform
Engineered a full-stack authoring platform powered by a custom meta-framework, which reduced the human effort for creating interactive learning content by 93% (from 5 hours to 20 minutes).
Automating Evaluations: Cracking the Code of Subjective Grading with AI
A full-stack prototype for grading subjective answers with LLMs, built during my Agentic AI internship at USDC Global.
AI-Powered Automation of Academic Certificate Verification
An End-to-End Pipeline for Document Classification and Information Extraction, Reducing Verification Time by over 90%
Real-time Flight Tracking and Airspace Visualizer | Real Time Analytics
A Real-Time Flight Tracking Dashboard With Airspace Visualization Using Kafka and Superset
Multilingual Sentiment Analysis using LLMs
Finetuning a Llama 3.1 8b model to understand the sentiment of Indian Languages
Engineering a Scientific Data Platform for ISRO's AstroSat SSM Mission
A 14 TB data platform for ISRO's AstroSat SSM mission, letting researchers find, extract, and analyse observations themselves instead of requesting them.
VisuaLLMize
Easing the process of Exploratory Data Analysis using LLMs!
eLibrary: An Online Library
A fully featured online library application with Client and Admin RBAC, Caching, Bulk Mailing and Async Export Tasks
Taxi Fare Prediction
A extensive end-to-end machine learning pipeline to predict Taxi Fare using various qualitative and quantitative factors.
IITMBS Film Society Website & Blog
A community for all things films! Also a 'monthly' blog
Quantified Self
A Flask based web application to track your daily activities and habits.
No projects match these filters. Try another category or technology.