A full collection of AI/ML projects, research prototypes, and engineering work.
A hybrid tax assistant for Indian freelancers, students, creators, and small businesses that combines AI reasoning with deterministic tax logic, document extraction, and compliance insights.
A containerized real-time Store Intelligence API built with YOLOv8 nano, OpenCV, and FastAPI to track customer entry and exit, separate staff from shoppers, monitor queues, map movement across zones, and surface purchase conversion insights instantly.
An AI-powered calling agent that helps customers find products through natural phone conversations. Built with Twilio, ElevenLabs, Groq LLM, and ChromaDB with semantic search plus metadata filters.
Project preview
A network anomaly detection system using phishing data with modules for ingestion, transformation, training, logging, exception handling, and utility functions.
Project preview
Forecasts per-theatre daily audience attendance using booking and POS data. Includes temporal, holiday, and behavioral features, with handling for zero-demand and closure days.
A caption generator trained on Flickr8k that combines DenseNet201 feature extraction with an LSTM language model to produce short natural-language captions.
A custom Transformer-based model with an attention layer was implemented to capture contextual relationships within text sequences. The performance of the Transformer model was then compared with LSTM and GRU models by evaluating metrics.
An end-to-end computer vision project built on the CT-KIDNEY-DATASET-Normal-Cyst-Tumor-Stone. The workflow covered EDA, clean data preparation, image preprocessing, and CNN model training for kidney tumour detection.
An AI legal agent that answers law-related questions, analyzes legal documents, and helps surface practical legal strategies with a risk-aware perspective.
A simple agentic AI chatbot dedicated to finance and stock-related questions, designed to help with investment strategies, recent news, stock behavior, and analyst recommendations.
A customer churn prediction application built to identify retention risk and support better customer planning through model-driven predictions.
A machine learning app that predicts IPL match outcomes using historical patterns and match context, built for quick and practical win-probability exploration.
A movie recommender system that suggests similar titles and supports simple discovery flows for finding what to watch next.