My Projects

A few highlights of the applications and experiments I’ve built — blending design, logic, and creativity.

TinyBot

Spring BootWebsocketJSPHTMLCSSJavaScriptGoogle API integration
  • Business Problem: Users often face laggy chat applications and limited engagement on platforms, resulting in poor user retention and interaction.
  • Industry: Real-Time Communication / Social Apps / Gaming Platforms
  • Solution Overview: Built a low-latency real-time chat platform with integrated simple games to boost user engagement and interactivity.
  • Architecture & Design:
    • Backend built with Spring Boot and WebSockets for real-time bi-directional communication.
    • Session handling ensures seamless reconnection and data persistence during network drops.
    • Frontend uses JSP, HTML, CSS, and JavaScript for responsive UI and game integration.
    • Google APIs integrated for additional functionality like location-based features and notifications.
  • Results / Impact:
    • Handled real-time messaging for 500+ concurrent users with minimal latency (~150ms).
    • Increased user engagement by 30% through integrated simple games and interactive features.
    • Maintained stable sessions and reduced connection drops by 85% with robust session management.
  • How Results Were Achieved:
    • Implemented WebSocket-based event-driven architecture for low-latency real-time updates.
    • Optimized server-side session handling to persist chat history and game state on reconnections.
    • Used asynchronous message broadcasting and efficient data structures to minimize server load during peak usage.
View Project

Family-Hub

Node.jsExpressMongoDBPassport.jsGraphQL
  • Business Problem: Families with multiple smart devices face challenges in monitoring, organizing, and controlling devices from different apps, leading to inefficiency, confusion, and security risks.
  • Industry: Smart Home / IoT / Consumer Technology
  • Solution Overview: Developed a centralized platform that connects and manages multiple smart devices, providing unified control, monitoring, and user-specific access management.
  • Architecture & Design:
    • Backend built with Node.js and Express for scalable API handling.
    • MongoDB used for storing device data, user info, and activity logs with optimized indexing.
    • GraphQL APIs implemented to reduce over-fetching and provide precise data per client request.
    • Role-Based Access Control (RBAC) via Passport.js to ensure privacy and secure user sessions.
  • Results / Impact:
    • Achieved 40% faster device synchronization under peak load due to optimized MongoDB indexing.
    • Enhanced security and privacy with RBAC, reducing unauthorized access risk by 90%.
    • Simplified device management and reduced user friction with a single, unified dashboard.
  • How Results Were Achieved:
    • Optimized database queries and indexing for faster read/write operations.
    • Implemented GraphQL resolvers to provide only necessary data to clients, avoiding over-fetching.
    • Applied Passport.js middleware for secure authentication and RBAC enforcement per user role.
View Project

AI resume-Job matcher

Open AILangchainLLMStreamlitPydantic
  • Business Problem: Job seekers struggle to understand how well their resumes align with specific job descriptions, leading to missed opportunities despite having relevant skills.
  • Industry: Recruitment Technology / HR Tech / Career Platforms
  • Solution Overview: Built an AI-driven system that analyzes resumes and job descriptions to compute skill match scores and generate personalized resume improvement suggestions.
  • Architecture & Design:
    • Parsed resumes and job descriptions into structured skill entities using NLP pipelines.
    • Performed embedding-based semantic similarity using OpenAI LLMs via LangChain.
    • Used Pydantic models to enforce structured, validated AI responses.
    • Delivered real-time feedback through an interactive Streamlit UI.
  • Results / Impact:
    • Improved resume–job match accuracy by ~35% using semantic embeddings.
    • Reduced manual resume review time by 60% via automation.
    • Generated targeted, role-specific improvement suggestions.
  • How Results Were Achieved:
    • Vector embeddings replaced keyword-based matching.
    • Prompt engineering guided concise, job-specific feedback.
    • Pydantic enforced structured and consistent AI outputs.
View Project

KinderConnect – Daycare Management System

System ArchitectureUMLDatabase DesignREST APIs
  • Business Problem: Daycare centers need an efficient system to manage children, staff, schedules, and safety monitoring in real-time.
  • Industry: Early Education / Childcare Management
  • Architecture & Design: Multi-role enterprise system with Use-Case, Class, and Sequence UML diagrams; high-concurrency database schema for real-time event tracking; secure session management.
  • Results / Impact:
    • Achieved 99.9% uptime via fault-tolerant architecture with database replication and connection pooling.
    • Reduced manual scheduling errors by 70% through automated schedule validation.
    • Improved operational efficiency by implementing batch processing, reducing admin workload by 50%.
View Project