Manan Nagpal

Built a full-stack SaaS platform with Monaco Editor workspaces and drag-and-drop uploads for multiple programming languages. Engineered AST-based static linting with OpenAI-powered review for bug detection, refactor suggestions, and complexity auditing. Developed a FastAPI and PostgreSQL service layer with JWT authentication and API rate limiting.

Architected and deployed a full-stack AI platform using the Gemini API to parse resumes against job descriptions and surface skill-gap analysis. Designed structured prompt chains for ATS-optimized resumes and role-specific interview questions. Built a headless Puppeteer pipeline for server-side PDF generation, secured API routes with JWT authentication, and deployed on Vercel using modular MVC architecture.

Architected and developed ScopeAI, a production-ready AI-powered SaaS platform that automates code quality audits, detects security vulnerabilities, and evaluates code complexity by combining traditional static analysis with LLM-driven code reviews. Key Highlights Built a hybrid code analysis pipeline combining AST-based static linting with OpenAI-powered code reviews for bug detection, refactoring suggestions, and logic analysis. Developed an interactive Monaco Editor supporting multi-language code editing, drag-and-drop uploads, and line-level navigation to detected issues. Implemented code quality metrics including Cyclomatic Complexity, Maintainability Index, nesting depth, and lines-of-code analysis. Designed analytics dashboards with Recharts to visualize scan history, language distribution, and quality trends. Built a secure authentication system using JWT with refresh tokens and developed a scalable backend using Express, Prisma, and PostgreSQL. Tech Stack: React.js, Next.js 15, TypeScript, Node.js, Express.js, PostgreSQL, Prisma, Tailwind CSS, Monaco Editor, OpenAI API, Recharts

Prepify — AI-Powered Job Preparation System Built a full-stack MERN application implementing a custom Retrieval-Augmented Generation (RAG) pipeline for intelligent interview preparation and resume analysis. Developed transformer-based semantic search using all-MiniLM-L6-v2 embeddings and cosine similarity vector retrieval to generate context-aware interview responses using Groq-hosted LLMs. Implemented AI evaluation engine for scoring technical accuracy, communication, confidence, and clarity, along with resume PDF analysis and personalized interview question generation. Tech Stack: React.js, Tailwind CSS, Node.js, Express.js, MongoDB, Groq SDK, Transformer Embeddings, Semantic Search, JWT Authentication.

Built a Machine Learning-based Exoplanet Detection and Visualization system using NASA’s open-source light curve datasets. The system analyzes stellar brightness variations and identifies periodic dips in luminosity to predict the likelihood of exoplanet presence for a given star ID. Developed and tested ML workflows in Jupyter Notebook, including data preprocessing, model training, and light curve analysis. Built an interactive frontend using React.js and React Three Fiber to visualize planetary data through immersive 3D space scenes and responsive dashboards. Implemented real-time plotting, API-based data handling, and visualization tools to improve astronomical data interpretation and user interaction. Tech Stack: Python, Jupyter Notebook, Scikit-Learn, Matplotlib, Lightkurve, React.js, React Three Fiber, Three.js, JavaScript, REST APIs, NASA Exoplanet Archive