Featured Projects
A showcase of my technical projects, ranging from machine learning models to full-stack web applications. Click any card to view details.
Autonomous Multi-Agent Financial Research Platform
Designed and implemented a sophisticated multi-agent system that mimics the workflow of a real-world hedge fund. Instead of a single LLM hallucinating answers, this architecture assigns distinct roles to specialized agents that must debate, validate, and reach a consensus before making a trading recommendation.
Adversarial Robustness Evaluation for LLMs
Built an adversarial robustness pipeline using PyTorch, Transformers, and Hugging Face to test GPT-2 on 600+ prompt-attack-defense scenarios, achieving 99.06% classifier accuracy.
Smart Retail Ecosystem (Winner, ASU GlobeStrat'25)
AI-driven retail platform connecting store owners, CPG brands, and shoppers to modernize inventory management and customer experience.
Multivariate S&P 500 Forecasting Using News Sentiment and Global Indices
Engineered an LSTM forecasting model combining S&P 500, Nikkei, Hang Seng, Shanghai data with FinBERT sentiment to predict market trends.
Real-Time Phishing Detection using Graph Neural Networks (GNNs)
Developed a full-stack phishing detection system with GCN3, extracting lexical, domain, and HTML features for real-time browser protection.
Sustainable Innovation Web Application
Full-stack Next.js application promoting sustainable waste management using Google's Gemini AI for creative upcycling suggestions
Genre Odyssey: Film Visualization
Interactive D3.js visualization mapping relationships across genres, directors, and films
Adversarial Attacks on AI Models
CNN model to analyze the CIFAR-10 dataset
Oral Cancer Detection & Classification
DenseNet architecture achieving 95.24% accuracy in cancer classification
Bird Species Detection using CNN
CNN model for classifying bird species from audio recordings
Crop Price Prediction
Machine learning model for agricultural commodity price forecasting