DL ECG Monitoring System

A comprehensive IoT-based (ESP 32) cardiac monitoring system that captures ECG signals from patients, processes them through encrypted cloud pipelines, and provides real-time anomaly detection using deep learning to alert medical professionals of potential cardiac abnormalities.

Deep Learning, Embedded Systems, Python, IOT, Encryption, Networking, AWS

Vibration Analysis System

Built fault detection model with Random Forest and Signal Processing (SciPy), detecting failures 7 days earlier. Developed and deployed dashboard using Django, with CI/CD through GitHub Actions, pytest and AWS ECS.

Random Forest, Django, Signal Processing (SciPy), Numpy, GitHub Actions, AWS

Electricity Usage Prediction

Developed prediction model using XGBoost, log transformation and stratification, achieving 90% accuracy. Engineered data pipeline for 2M+ electricity records in Pandas to integrate temperature and population features.

XGBoost, Pandas, NumPy, Scikit-learn, Vercel

Ascend AI

Developed AI-powered Habit Tracker using TypeScript, React.js, MongoDB, currently 70+ users on waitlist. Implemented RAG to Atomic Habits using LangChain and AWS Bedrock API, generating personalized habit plans

React JS, LangChain, Pinecone, AWS, Clerk, MongoDB

Rate My Professor Assistant

Built intelligent review assistant using Web Scraping, Node.js and Pinecone DB, covering 8000+ universities. Implemented RAG pipeline using LangChain and OpenAI API for user queries, achieving 95% accuracy

Next JS, React JS, OpenAI, LangChain, Pinecone

Flashcard SaaS AI

Developed Next.js app for customized study materials using Llama 3.1 API, generating 50+ flashcards at a time. Integrated Stripe for payments and Clerk for authentication, securing 30+ user sign-ups

Next JS, React JS, Gemini, Stripe, Clerk, Firebase

Pantry Tracker AI

Built smart inventory system using Next.js and Firebase with CRUD functionality, optimizing grocery tracking. Integrated Gemini API to generate recipes from soon-to-expire ingredients, reducing food waste by 20%

Next JS, React JS, Gemini, Firebase

Customer Support AI

Worked in a team of 3 to develop an interactive customer support agent for Harvard IT Department, integrated a custom RAG pipeline using Gemini and Pinecone that responds based on knowledge base articles

Next JS, React JS, Gemini, LangChain, Pinecone, Python

Reasoning Abilities Agent

Created an intelligent agent capable of breaking down complex queries into subtasks, reasoning about how to complete each subtask, executing the necessary actions, and evaluating the results. This agent leverages OpenAI's language models to perform these cognitive functions.

Python, Prompt Engineering, OpenAI

BeeSweeper Game

Developed in JavaFX, pluck flowers using numerical clues to identify safe flowers and avoid bees by placing beehives on suspect flowers

Enhanced strategic thinking and problem solving skills

Java, JavaFX

Store Checkout Simulator

Developed a Java program which optimized customer service time by simulating the checkout process.

Monitored customer arrival times and assigned them to the shortest line, aimed at reducing average wait times

Java