Projects
Projects
🚀 Web & Cloud Solutions
- Designed and deployed scalable e-commerce platforms using AWS, Terraform, and microservices.
- Built cloud-based applications with auto-scaling, CI/CD pipelines, and high availability.
- Developed secure REST APIs and GraphQL services for efficient data communication.
- Created a Meeting Notes Application with vanilla JavaScript, featuring CRUD operations, dynamic action item management, and smooth CSS transitions.
📊 Data & GIS Applications
- Engineered Java-based GIS solutions for spatial data analysis and visualization.
- Implemented data-driven decision-making tools using PostgreSQL, MongoDB, and ElasticSearch.
- Created automated ETL pipelines for large-scale data processing and analytics.
🔗 Blockchain & FinTech
- Developed crypto-based payment platforms like Pay2Fix, enabling micropayments via smart contracts.
- Designed Ethereum-based DeFi solutions with secure wallet integrations and tokenized assets.
🤖 AI & Machine Learning
- Built an AI-powered Wyckoff Trading Assistant using PyTorch Transformer neural networks trained on 1,187 Q&A pairs for market analysis.
- Implemented reinforcement learning algorithms (Q-learning) for automated trading strategies with backtesting on historical stock data.
- Developed a computer vision-powered control system for the Kinova Robotic Arm with path-planning algorithms.
- Optimized sensor fusion techniques and robotic path-planning for autonomous navigation.
MERN Stack
ElderlyCare
A web-based solution to improve healthcare accessibility for the elderly, EC-Buddy AI assistant, book doctor appointments, and empower doctors with real-time access, enhancing operational efficiency and patient satisfaction.
Blockchain, Frontend
Secure Social Share
An E2E encrypted data sharing system with strict, smart contract-based access control, allowing users to securely share sensitive information with their social circle in a decentralized way. Winner of multiple 1st place prizes at ETHBoston 2024.
Machine Learning, Backend
AI-Powered Richard Wyckoff Trading Assistant
Wyckoff Trading Assistant is an intelligent web application that combines a Transformer-based AI chatbot with reinforcement learning to help traders analyze stocks using Richard Wyckoff's methodology. The system features a PyTorch-powered neural network trained on 1,187 Wyckoff Q&A pairs to answer questions about market structure, accumulation, distribution, springs, and upthrusts, while a Q-learning algorithm learns optimal trading strategies through backtesting on historical data. Built with Flask, the application fetches real-time stock data via yfinance, enriches it with technical indicators (Moving Averages, RSI, Bollinger Bands), and visualizes results through an interactive Chart.js dashboard, providing both educational insights into Wyckoff principles and practical algorithmic trading analysis with customizable backtesting parameters.
AWS, Terraform
Cloud Native Application
Developed a cloud-native user management application using Node.js, Express.js, and PostgreSQL, supporting concurrent requests, CRUD operations, health monitoring, and robust backend services with high availability. Provisioned AWS infrastructure using Terraform and Packer, deploying VPCs, RDS, Route 53, and SSL certificates. Integrated AWS SNS and Lambda for event-driven workflows, including Mailgun email verification. Secured services with Systemd and network security groups. Enhanced scalability with EC2 Auto Scaling and Elastic Load Balancing, enabling faster deployments. Implemented a CI/CD pipeline via GitHub Actions, automating testing, AMI creation, and rolling updates for seamless deployment.
Vanilla JavaScript, SCSS
Meeting Notes Application
A simple yet powerful meeting notes application built with vanilla JavaScript. Features include creating notes with titles, content, and action items, viewing all notes with truncated content (10 words max), expanding notes with smooth CSS transitions to see full details, editing content and action items in expanded view, and toggling action items between open and completed states using checkboxes. The application fetches existing notes from a JSON file using XMLHttpRequest and allows users to add new notes with automatically generated creation timestamps. Built with modular SCSS for styling and a clean, responsive interface.
📈🤖
RockyBot
Financial News Research
Python, LangChain, OpenAI, FAISS, Streamlit
RockyBot - News Research Tool
A user-friendly news research tool for the stock market and financial domain. Users can input article URLs and ask questions to receive relevant insights. Features include loading URLs to fetch article content, processing through LangChain's UnstructuredURL Loader, constructing embedding vectors using OpenAI's embeddings, and leveraging FAISS for swift similarity search retrieval. Interact with ChatGPT to get answers along with source URLs.