Explore my work and creative solutions
A dual-modal AI platform using EfficientNetB0 (faces) and HuBERT (speech) to help autistic individuals understand emotions. Features real-time webcam capture, top 3 emotion charts, tips, reports, and a modern HTML/CSS/JS interface with dark/light themes.With Botpress conversational chatbot integration.
Enterprise-grade phishing detection system designed to identify malicious emails and URLs using advanced ensemble machine learning models, TF-IDF–based NLP, and context-aware hybrid fusion techniques. The platform delivers real-time threat analysis through a Flask-powered web application and Chrome extension, enhanced with confidence scoring and intelligent anomaly detection for improved security insights.
Automates attendance tracking using real-time facial recognition powered by OpenCV and scikit-learn, combined with a user-friendly Streamlit web interface. The system captures and processes live video feeds to accurately identify individuals, mark attendance instantly, and generate organized reports for efficient monitoring and analysis.
A comprehensive web application that showcases local businesses and attractions across Bangalore city.
A comprehensive web application for heart disease risk assessment and management using machine learning.
Helps farmers determine the most suitable crop for cultivation based on soil and environmental conditions.
A web application that allows users to upload a recipe image, which is then analyzed by a trained machine learning model to predict the recipe.
An intelligent machine learning system that analyzes news articles using NLP techniques such as TF-IDF and supervised classification models to accurately distinguish between real and fake news, improving information credibility and media trust.
A responsive front-end replica of the Apple Store built using React, focusing on modern UI/UX design principles, component-based architecture, smooth interactions, and pixel-perfect layout implementation for real-world e-commerce interfaces.
A beautifully designed restaurant website "La Pizzeria Elegante" showcasing elegant menus and layouts.
Classifies brain MRI images to detect tumors using machine learning models.
A modern, AI-powered writing assistant built with Next.js that uses OpenAI's API to enhance or auto-generate content.
A futuristic React application that allows users to create and customize their own avatars with various options like face shape, skin tone, hair style, eyes, clothes, and more.
Classifies waste materials (cardboard, glass, metal(cans only), paper, plastic) from images using MobileNetV2 and Focal Loss.
A modern, AI-powered weather app built with Next.js 15 that gives real-time weather info, clothing/activity suggestions, and elegant UI with dark/light mode.
AI-powered virtual assistant built using Voiceflow to handle customer queries, recommend services. Integrated RelevanceAI for price prediction, Make.com for automation with Google Sheets, and deployed via Flask.
A full-stack automated quiz and test platform built with Flask. Features include secure authentication, role-based dashboards, timer-based assessments, intelligent grading with fuzzy/semantic matching, interactive analytics, CSV export, and admin statistics.
A deep learning-based face mask detection system built using a custom CNN. It includes preprocessing, data augmentation, and a 3-layer architecture with batch normalization and dropout for strong generalization. Achieved 96.82% accuracy and 0.9965 AUC-ROC with full evaluation. Now deployed as a stunning interactive web app via Streamlit.
A full-stack platform for cybersecurity and data science aspirants. Features AI-driven mock interviews, intelligent resume analysis, real-time feedback, performance analytics, JWT-based authentication, and exportable reports.
A full-stack NLP application that analyzes movie reviews using TextBlob, VADER & DistilBERT for sentiment and spam detection, with YouTube comment analysis, OMDB integration, dashboards, and a modern responsive UI.
A high-performance computing project comparing serial and OpenMP-parallel image processing pipelines in C, featuring cycle-level benchmarking, JSON-based metrics logging, and a Flask + Chart.js dashboard for interactive performance visualization.
A modern AI-powered platform that generates engaging social media captions and optimized hashtags from uploaded images. Built with React, TypeScript, and Supabase, it integrates computer vision, tone-based caption generation, secure authentication, usage credits, and Razorpay-powered Pro upgrades.
An enterprise-grade transit intelligence platform that predicts public transport delays using a physics-informed ensemble ML model with a Claude AI reasoning layer and real-time 3D visualizations. Built with FastAPI and React, it delivers sub-3 second predictions with risk scores, confidence intervals, SLA alerts, and optimal departure insights. Includes a 3D transit map, heatmaps, and live input monitoring.
An enterprise-grade AI document intelligence platform that transforms raw documents into actionable insights through context-aware chat, semantic search, and automated analysis. Built with React and Node.js, it integrates Google Gemini for grounded Q&A, vision-based image understanding, and executive summaries. Features include interactive CSV analytics dashboards, named entity recognition with knowledge graphs, document comparison, and secure Supabase authentication for scalable, real-world usage.
A production-grade real-time multi-object tracking platform for intelligent surveillance and traffic analytics. Built with FastAPI and React, it leverages YOLOv8 and ByteTrack for accurate object detection and persistent identity tracking across live video streams. Features include real-time webcam and RTSP streaming, anomaly detection (loitering, wrong-way, density), line-crossing analytics, and interactive dashboards with live metrics and CSV export, deployed on scalable CPU-based infrastructure without GPU dependency.
An AI-powered conversational data intelligence platform that lets you chat with an LLM, upload CSV datasets, and ask questions in plain English to automatically generate and execute pandas analysis code. Built with Streamlit and LangChain, it features AST-based safe code execution, auto-generated Plotly visualizations, one-click AI insight generation, persistent conversation memory, and a fully custom dark-themed UI — all powered by Groq's ultra-fast Llama 3.3 70B inference engine.
A conversational data intelligence platform that dynamically converts any uploaded CSV into a SQLite database and lets you interrogate it using plain English, powered by a LangChain ReAct agent with self-correcting SQL generation. Features auto-selected Plotly visualizations, dual API key support for local and cloud deployment, query history replay, CSV export, and a premium dark editorial UI — all running on Groq's ultra-low-latency Llama 3.3 70B and Mixtral inference.
A fully automated AI content engine built on n8n that fetches trending AI news via SerpAPI, generates complete blog articles using GROQ's Llama 3.3 70B, and delivers professionally formatted HTML newsletters to your inbox daily — with zero manual effort. Features a 6-node cloud-hosted pipeline with cron scheduling, JavaScript-based article extraction, LLM prompt engineering for structured HTML output, and Gmail delivery via App Password authentication.
An enterprise-grade autonomous AI research platform inspired by Perplexity AI that performs multi-step research, intelligent web browsing, source extraction, and AI-powered report synthesis with citations. Built using LangGraph orchestration, Claude AI reasoning, FastAPI backend services, Playwright-based web automation, Redis task handling, and a responsive Next.js frontend featuring real-time progress tracking, interactive knowledge graphs, confidence analysis dashboards, and exportable research reports.
An enterprise-grade ESG intelligence platform that automates corporate ESG scoring, greenwashing risk detection, controversy analysis, and institutional portfolio evaluation using AI-powered financial analytics. Built using FastAPI backend services, Claude AI reasoning, semantic vector search, and a responsive React frontend featuring interactive ESG dashboards, SEC filing analysis, portfolio stress testing, controversy timelines, and exportable institutional-grade financial reports.
An end-to-end machine learning platform for high-precision hourly building electricity consumption forecasting using authentic UCI sensor data. Built with XGBoost, FastAPI, React, and TypeScript, the platform provides explainable predictions using TreeSHAP, empirical 90% confidence intervals, out-of-distribution reliability detection, live weather integration, what-if energy simulations, actual-vs-predicted tracking, and automated energy-saving recommendations with production deployment across Vercel and Render.