Agentic AI-Powered Amazon Product Scraper

Designed and developed a product scraper integrated with Large Language Models (LLMs) to autonomously extract, format, and present Amazon product data using agentic workflows. Created an interactive Streamlit UI that enables real-time product search, dynamic filtering, and result visualization.

Python LLMs Streamlit Web Scraping
100
Autonomous
95
Accuracy %
GitHub

Real Estate Price Predictor

Enhanced predictive modeling with 94% accuracy using XGBoost regression model for housing prices. Created a location-driven content-based recommender system for personalized user experience.

XGBoost Machine Learning Streamlit Python
94
Accuracy %
1000
Predictions
Live App GitHub

News Summarization and Hindi TTS App

Built a scalable NLP pipeline using Hugging Face Transformers, resulting in a 30% reduction in processing time for article summarization, sentiment analysis, and Hindi TTS output. Achieved 92% ROUGE-L score and 87% sentiment classification accuracy.

NLP Transformers TTS Sentiment Analysis
92
ROUGE-L %
30
Speed Boost %
GitHub

Image Compression with Autoencoder

Built CNN-based autoencoders in PyTorch to compress images by 60% with less than 10% visual loss. Achieved 0.91 SSIM score and 20% MSE reduction over baseline, demonstrating efficient lossy compression with minimal quality degradation.

PyTorch CNN Autoencoders Computer Vision
60
Compression %
91
SSIM Score
GitHub