Projects 💻

Step into my laboratory of innovation, where comfort zones are merely starting points and each project is a bold leap into the unknown—here's to coding daringly and living on the edge of possibility!

Real-Time Document Collaboration Service | Go, gRPC, WebSocket, AWS, Docker, Protocol Buffers | GitHub

• • Developed a real-time document editing platform with WebSocket communication for live updates and user synchronization
• • Implemented session management and document versioning, ensuring consistent data integrity across multiple users
• • Enhanced user experience by integrating gRPC and Protocol Buffers for efficient communication and DynamoDB for robust data storage
• • Utilized AWS EC2 for scalable deployment and Docker for containerization, ensuring high availability and portability of the service
• • Designed the system to handle concurrent user edits, minimizing conflicts and providing real-time conflict resolution
• • Improved document synchronization speed by 20% through optimized WebSocket handling and efficient data transmission protocols

StreamCraft - Video Streaming Service | Javascript, AWS (S3, DynamoDB, Lamda, CloudFront), Kafka | GitHub

• • Increased concurrent upload capacity by 30% & reduced processing time, by integrating AWS SQS for efficient uploads to S3
• • Reduced data sync time by 15% using DynamoDB and Lambda, enhancing event-driven triggers & real-time handling
• • Introduced real-time alerts for video creators via Kafka, thus increasing user engagement in terms of comments and likes
• • Enhanced video playback quality by coupling AWS S3 with CloudFront, ensuring high-quality content delivery to end-users

Job Description Summarization Plugin | Flask, Generative AI, Large Langauge Models, AWS

• • Developed browser plugin for summarizing job descriptions & identifying key skills, using Falcon 7B (Large Language Model)
• • Employed Flask for LLM interactions and Node.js with Express.js for managing other backend services, hosted on AWS
• • Enhanced plugin scalability and reliability by using AWS EC2 for hosting and AWS Lambda for asynchronous operations

CROWD - A Social Network Application | MongoDB, Express.js, React, Node.js (MERN) | GitHub

• • Built full-stack social media application using MERN stack, enabling user registration and login, & dynamic real-time updates
• • Designed a responsive and engaging frontend with ReactJS, integrated with an Express.js API for user activity handling
• • Utilized MongoDB’s NoSQL flexibility for efficient, scalable management and quick retrieval of user data, posts, and comments

Stock Portfolio Management | Java, MVC, SQL, Object-Oriented Programming & Design | GitHub

• • Developed a real-time stock portfolio management application in Java, achieving robust and scalable design by implementing MVC architecture with design patterns like Factory and Singleton, adhering to SOLID principles
• • Attained 97% code coverage and minimized potential defects by rigorously writing JUnit tests for functionality testing
• • Ensured efficient data caching and accessibility by fetching real-time stock data using the Alphavantage API, storing it in SQL, and employing programming constructs like stored procedures, functions, and triggers

Credit Worthiness of a Customer | Self-supervised Machine Learning, Binary Classification Model | GitHub

• • Increased concurrent upload capacity by 30% & reduced processing time, by integrating AWS SQS for efficient uploads to S3
• • Reduced data sync time by 15% using DynamoDB and Lambda, enhancing event-driven triggers & real-time handling
• • Introduced real-time alerts for video creators via Kafka, thus increasing user engagement in terms of comments and likes
• • Enhanced video playback quality by coupling AWS S3 with CloudFront, ensuring high-quality content delivery to end-users

Face Mask Detection and Face Recognition | Computer Vision, Deep Learning | GitHub | Manuscript

• • Optimized ResNet101 architecture to accurately detect face masks over faces, achieving a remarkable 99% detection accuracy
• • Implemented a Deep Metric Learning network with ResNet-34 architecture, for recognizing individuals’ faces with precision

Tool for Automatic Question Generation | Natural Language Processing, NLTK, SpaCy | GitHub | Manuscript

• • Attained 87% accuracy in identifying answer phrases using a Gaussian Naive Bayes classifier on the SQuAD 1.0 dataset
• • Utilized word embeddings and cosine similarity to generate distractors for the questions, enhancing the tool’s robustness