A collection of engineering projects spanning AI/ML, Blockchain, and Full-Stack development.
Fintel is an AI-powered financial intelligence mobile app designed to eliminate manual expense tracking by automatically reading UPI transaction SMS notifications in real-time. Utilizing a team of specialized agents, it classifies transactions, analyzes cashflow, and optimizes budgets while keeping data encrypted locally via SQLCipher. The platform extends into an investment engine for the Indian market, unifying portfolio analysis, mutual fund evaluation, and SEBI-aware risk management through an MCP-driven agentic tool selection layer. https://www.canva.com/design/DAG5_nGzA9E/xjAos89R6OImdoqZItYeEg/view
Developed a decentralised, AI-powered platform for secure document storage, categorisation, digital signature, and natural language-based document interaction.
Developed a multi-functional legal AI assistant for document creation, simplification & IPC chatbot.
Utilised machine learning techniques for portfolio optimization, considering risk tolerance and historical data. Built a user-friendly interface (playground) for easy portfolio construction, rebalancing and visualization. Post Hackathon, I have kept updating the project with new features and optimizations.
A.I Resume Parsing & Candidate Ranking System. Built AI-powered HR automation for resume parsing using OpenAI & NER models.
India is the second-largest food producer, yet wastes 20% of its food annually. FoodCheckAI is a "Phygital" solution that tracks household food quality and connects surplus restaurant food with NGOs to minimize wastage.
Developed a Retrieval-Augmented Generation (RAG) application focused on the Bhagavad Gita, utilizing a comprehensive dataset encompassing all chapters in Sanskrit, English, and Hindi. The system enables semantic search and conversational interaction with ancient scriptures, providing contextually grounded responses.
Developed a decentralized application (dApp) designed to bring transparency and traceability to the construction material supply chain. Built with Solidity smart contracts on the Ethereum blockchain, the platform enables real-time tracking of material lifecycles—from supplier to construction site—while ensuring data integrity and secure interactions via MetaMask and Ethers.js.
Developed Dzillow, a decentralized real-estate marketplace built on the Core Chain blockchain. The platform utilizes Solidity for smart contract development and IPFS for decentralized storage of property metadata, ensuring a transparent, secure, and tamper-proof environment for real estate listings and transactions.
Advanced retirement portfolio optimizer leveraging skfolio's sophisticated ML algorithms for multi-objective portfolio optimization. Features 8+ optimization strategies (Mean-Risk, HRP, NCO, DRO CVaR, Stacking, Benchmark Tracking), efficient frontier visualization with interactive parameter tuning, walk-forward backtesting for out-of-sample validation, target-date/glide-path planning with automatic rebalancing, risk metrics (Sharpe, Sortino, Calmar ratios, CVaR), portfolio clustering analysis, and Monte Carlo simulations. Supports Indian NSE stocks with customizable constraints and risk tolerance-based allocation generation. Includes comprehensive retirement projection and goal achievement analysis.
Built an interactive Streamlit application for real-time stock sentiment analysis. Fetches financial news from Finviz using BeautifulSoup, analyzes sentiment using NLTK's VADER analyzer on headlines and descriptions, visualizes trends with Plotly, and displays historical stock price data using yfinance. Features multi-stock selection, customizable update intervals, and detailed sentiment metrics with interactive charts.
Interactive real-time stock technical analysis dashboard with advanced data processing capabilities. Supports multiple chart types (Candlestick, Line), multiple timeframes (1D, 1WK, 1MO, 1Y, Max), and technical indicators (SMA 20, EMA 20). Features robust data handling with MultiIndex column flattening, timezone-aware datetime processing, and error recovery. Displays comprehensive metrics (OHLCV), real-time prices for 45+ Indian NSE stocks, historical data tables, indicator analysis with comparative explanations. Uses Plotly for interactive visualizations and automated data validation to ensure data integrity.
Engineered a real-time data pipeline to visualize Zomato (ZOMATO.NS) stock data. The architecture leverages Python for data ingestion, Apache Kafka for distributed messaging, Apache Druid for sub-second OLAP queries, and Apache Superset for live dashboarding and technical analysis.
An end-to-end algorithmic trading research/demo platform combining interactive Streamlit dashboards, rule-based trading strategies and a simple ML-based next-day price predictor. Key features: - Interactive Streamlit apps (streamlit_app.py and main_app.py) with tabs for Price Prediction and Trading Dashboard. - Prediction: loads a pre-trained logistic regression model and scaler from models/, builds a 50+ feature vector (technical, volatility, momentum, lag and volume features), scales it and predicts next-day up/down with a confidence gauge and model metrics shown in the sidebar. - Trading Dashboard: SMA/EMA rule-based strategies, configurable parameters (short/long periods, RSI period), risk-management options (transaction cost, stop-loss, take-profit), walk-forward/backtest UI and downloadable CSVs. - Indicators and features: computes RSI, SMA, EMA, MACD, Bollinger Bands and many engineered features (volatility windows, return lags, volume ratios, price position, etc.). - Visualizations: comprehensive Plotly charts — price with signals, equity curve vs buy-and-hold, RSI, MACD, Bollinger Bands, drawdown chart, trade timelines and distributions, and performance metrics (Final Value, Total Return, Win Rate, Sharpe, Max Drawdown, Volatility). - Backtesting: utilities for applying signals and backtesting (utils.backtester), producing per-trade details and summary metrics. - Indian market focus: predefined list of NSE tickers (.NS) supported throughout the UI. - Developer convenience: repository includes a PowerShell scaffolding script to create expected project structure (indicators/, models/, strategy/, utils/). - Usage: run CLI backtests with main.py or launch interactive UIs with `streamlit run streamlit_app.py` (prediction + dashboard) or `streamlit run main_app.py` (dashboard). Notes: the apps import modules from indicators/, strategy/ and utils/ and expect model files under models/ (e.g. logistic_regression_model.pkl and scaler.pkl).