Projects & Research
A collection of my research work, side projects, and technical explorations in AI, ML, and software development.
Technical Expertise
Backend & Infrastructure
AI & Machine Learning
Frontend & Tools
Smart TV Content Recommender and Assistant
A modern web application that uses Google's Gemini AI to enhance the Smart TV user experience. Features include personalized content recommendations based on viewing history, an interactive content explorer for show/movie information queries, and a natural language feature guide for TV settings. Built with TypeScript (75.3% of codebase) and React frontend, Node.js/Express backend.
View Project
AI/ML NewsFlow: Financial Market News Causal Analysis Platform
A financial news analysis platform that leverages spaCy, DBSCAN clustering and graph databases to identify and visualize causal relationships in financial markets. The system processes news from 30+ financial sources, extracts key information, and builds an interactive knowledge graph to reveal market dynamics.
View ProjectICD-5 Medical Code MatchingAI Engine
An medical code search system that leverages faiss's squared Euclidean (L2) distance and embedding models to automatically identify and match the user's query to the top 10 relevant medical codes. The International Classification of Diseases (ICD) is a system that uses codes to map health conditions to generic categories and their variations. These codes can be up to six characters long
View Project
Instant Stocks Query and Analytics in Various Currency Engine
A real-time stock market analysis that provides instant price query in different currency and Q&A capabilities. Utilizes AWS Bedrock for advanced natural language processing and market data analysis.
View ProjectAI Meeting Minutes Generator
An automated system that generates comprehensive meeting minutes from audio recordings. Leverages Google Gemini for advanced speech recognition and natural language summarization.
View ProjectMultimodal AI Query System
A web application that combines vision and text AI capabilities to answer queries about images and documents. Features include OCR processing, semantic search, and multi-modal reasoning using state-of-the-art AI models.
View Project
Research Paper Vector Search Q&A
An intelligent document search system that converts research papers into vector embeddings for semantic search. Integrates local LLMs with vector databases to enable natural language querying of technical documents.
View Project
Interactive Image Generation with Diffusion
A local deployment of Stable Diffusion 2.1 for interactive image generation. Implements custom prompt engineering and image processing pipelines for enhanced generation quality.
View Project
Text-to-SQL Assistant
An intelligent SQL query generator that translates natural language questions into SQL commands. Features include query visualization, CSV export, and interactive chart generation using Plotly.
View Project
QLoRA-SQLBase Model
A fine-tuned language model for text-to-SQL tasks using quantized LoRA (QLoRA) on Gemma 2b. Optimized for memory efficiency while maintaining high performance on SQL generation tasks.
View Project
Go-based AI API Service
A lightweight REST API service built in Go that provides a clean interface to Cohere's AI models. Features robust error handling, request validation, and containerized deployment.
View Project
Heart Disease Classification
A comparative study of machine learning models for heart disease prediction. Implements both traditional (Decision Trees) and deep learning (ANN) approaches with comprehensive performance analysis.
View Project
MNIST CNN Classifier
A custom CNN implementation for MNIST digit classification achieving 99.17% accuracy. Features a carefully designed architecture with three convolutional layers and advanced regularization techniques.
View ProjectVoting-Web-App
This voting web app allows users to vote on their favorite programming languages and view the results in real-time.
View Project