Projects
Resolved
Hackathon winning multimodal complaint analysis platform for a fintech startup
Key Highlights:
Integrated Google Cloud APIs, AssemblyAI, and OpenAI for analysis, summarization, and classification
Built an AI chat interface that delivered tailored insights using Pinecone-powered contextual retrieval across 1,000+ multimodal cases
Reduced processing time by over 70%, streamlining workflows and accelerating resolution
GhostLink
Paid platform for content creators to boost interaction and growth on LinkedIn
Key Highlights:
AI-driven video-to-text workflow for automating transcription and post creation
RAG pipeline guided by top-performing LinkedIn posts (50K+ impressions, 500+ reactions) to enhance engagement and align with creator success patterns
Collaborated with creators (totaling 1.5M+ followers) and refined through user feedback
Gym Haven
Online storefront built using Node.js & JavaScript
Key Highlights:
Developed 20+ core REST APIs and tested functionality with Jasmine suites
Refined the MongoDB database to optimize handling of over 100 user profiles and 500 product listings
Utilized OAuth 2.0, Stripe, OpenAI, EasyPost, Web3Forms APIs
DataSphere
Applied machine learning to solve real-world problems across finance, healthcare, and energy domains
Key Highlights:
Predicted S&P 500 stock returns using Lasso and Elastic Net regression on time-series financial data
Classified brain tumors from MRI scans using SVM and Random Forest, with image preprocessing across 5K+ training samples and 1K+ test samples
Uncovered global energy usage patterns using K-Means and K-Medians clustering across 5-year periods
RU Gyms Crowd Meter
Hackathon winning Python health application
Key Highlights:
1st place in health track out of 170 participants
Engineered a Python web app to display real-time gym crowd levels, designed to serve 50K+ students and faculty based on live user input
Deployed on a Linux server via an AWS EC2 instance
Serenity Sphere
Interactive chat avenues focused on holistic health
Key Highlights:
Omnichannel structure supporting 5 distinct AI chatbots and 1 live support option
RAG pipeline sourcing information from 40+ hand-picked YouTube videos
Established a PostgreSQL database to enable complex querying for improved chat capability
InShort
Personalized news curated from over 125 million online sources
Key Highlights:
Automated news extraction using Python and BeautifulSoup for content retrieval from GDELT Event Database
Implemented a RAG pipeline to deliver contextual insights from Pinecone DB based on user input
Waitlisted over 35 users within the first 10 hours post-launch
Rutgers Pizza
Interactive Android App to emulate a Pizzeria
Key Highlights:
Factory and Singleton design patterns
XML to customize seperate UIs for customers and the restaurant