Ankit Kumar

Built a full-stack AI mock interview platform where users upload a resume, select a target role, and complete personalized interview sessions with role-aware question generation. After each session, the system returns structured feedback with scores, strengths, weaknesses, and actionable improvement suggestions. Developed end-to-end with React.js, Node.js, Express.js, MongoDB, and TypeScript, integrating Gemini/Groq APIs for generation and evaluation. A key focus was reliability: I implemented prompt engineering, strict JSON response contracts, validation layers, retry logic, and optimized async workflows to handle inconsistent model outputs and deliver stable, production-like interview practice.

HRMS is a full-stack human resource management platform built with Next.js, Express, Prisma, PostgreSQL, Socket.IO, and Gemini AI. It brings core HR workflows into one role-based web application: employee records, attendance, leave, payroll, recruitment, performance management, announcements, notifications, reports, and an AI HR assistant. The project is designed as a portfolio-grade HR product and demonstrates typed API development, relational data modeling, role-based access control, real-time event delivery, LLM tool use, responsive frontend work, and practical validation tooling.

Advance-Weather-Station is a feature-rich full-stack weather application utilizing React, Next.js, and TypeScript. It provides real-time weather updates via the OpenWeatherMap API, supports geolocation, voice assistant, dynamic maps, and AI-powered weather tips. The app delivers an engaging, mobile-first user experience through responsive design, weather-based theming, interactive charts, and performance enhancements.

Built a full-stack hospital management web app with secure JWT cookie auth and role-based access for Patient, Doctor, and Admin. Features include patient registration/login, doctor listing with department-based selection, complete appointment APIs (create/get/update/delete + status), contact/message submission, admin-only doctor and appointment/message management, and Cloudinary-based doctor avatar uploads. Stack: React 19 + Vite 7 frontend, Node.js/Express + MongoDB backend.

Designed and developed a cultural heritage platform focused on traditional handloom and weaving crafts. The website features a clean multi-section layout with navigation for Home, About, Articles, and Contact, plus an admin access flow for content management. Built with a responsive UI to ensure a smooth experience across devices while highlighting stories, visuals, and educational content about handloom traditions.