Software Engineer | Full-Stack Developer (MERN) | Machine Learning & Blockchain | Python | AWS | Infosys Springboard AI

leetcode

Saket Chaudhary  • Peerlist

<h3>AI-Powered Mock Interview Platform</h3>
This is a comprehensive, production-ready, full-stack interview preparation marketplace. It connects candidates with expert interviewers, offering a seamless experience powered by real-time AI. Key features include:
<ul>
<li>Live HD Video Interviews with screen sharing and persistent chat.</li>
<li>AI-Generated Feedback Reports powered by Google Gemini.</li>
<li>Real-time AI Question Generator for interviewers during live calls.</li>
<li>Credit-based Booking System with subscription tiers.</li>
<li>Interviewer Earnings Dashboard with payout management.</li>
<li>Automated Email Notifications via Resend &amp; React Email.</li>
</ul>
The platform utilizes a robust tech stack including Next.js, Stream SDK for video and chat, Google Gemini API for AI capabilities, Clerk for authentication and billing, and Arcjet for security. It is deployed on Vercel.

Prept – AI-Powered Interview Marketplace Plat

Developed an end-to-end AI-based health prediction system aligned with UN SDG 3 – Good Health & Well-Being, designed to make healthcare insights more accessible and intelligent. Built using Machine Learning + Generative AI, the platform predicts diseases from symptoms, provides personalized health insights, and enables real-time AI-driven medical assistance. 🔍 Key Highlights: • Built a Random Forest model trained on 4,920 clinical samples, achieving 100% accuracy across 41 diseases & 131 symptoms • Designed an intelligent symptom checker with top-5 disease predictions and confidence scores • Integrated Google Gemini AI (Dr. AI chatbot) for real-time health queries, diet suggestions, and explanations • Developed a personalized health risk scoring system with interactive radar chart visualization • Created a medicine information module providing dosage, uses, and safety insights • Built a Streamlit web application with dashboards, analytics, and downloadable health reports • Implemented data preprocessing, feature engineering (one-hot encoding), and model comparison pipeline • Designed a scalable and user-friendly interface for better healthcare accessibility

🏥 AI-Powered Disease Prediction & Health Ass

🤖 AlphaAI - NLP-Powered Conversational Chatbot AlphaAI is a Machine Learning-based chatbot designed using Natural Language Processing (NLP). It predicts user intent and responds dynamically through a Streamlit web interface. The chatbot logs all interactions, providing a seamless and interactive experience. 💡 Key Features: ✔️ Intent Recognition – Uses TF-IDF & Logistic Regression for classification ✔️ Dynamic Responses – Generates predefined replies based on detected intent ✔️ Conversation Logging – Stores user interactions in chat_log.csv ✔️ Streamlit UI – Provides an elegant, user-friendly chat interface 🛠 Technology Used: NLP: NLTK, Scikit-Learn Web Development: Python, Streamlit Logging & Storage: CSV-based chat logs 🔹 Future Enhancements: Integration with BERT/GPT for more natural conversations Speech-to-text & text-to-speech capabilities Persistent database storage for conversation history

AlphaAI - Intelligent Chatbot

Chatty is a real-time full stack chat application built using the MERN stack (MongoDB, Express, React, Node.js), with integrated Socket.IO for live messaging and Cloudinary for image handling. ✨ Key Features: Real-time one-on-one messaging (text & image) Online/offline user status tracking Secure authentication (JWT, bcrypt) Profile management with Cloudinary 32+ dynamic UI themes (Tailwind CSS + DaisyUI) State management with Zustand Fully responsive UI and deployed using Render This project helped me dive deeper into real-time applications, state management, and deployment workflows in a production-like environment.

Chatty – Real-Time Full Stack Chat App

CodeCraft is a full-stack web development platform designed for practicing and showcasing coding projects. It allows users to explore code snippets, manage their profiles, and interact with a clean, responsive UI. The platform categorizes projects (Web, AI/ML, Blockchain, Data Science) and integrates features like dynamic routing, hover effects for tech stack previews, and a polished portfolio-style interface. Key Features: Dynamic project showcase with live GitHub integration Clean, responsive UI built with TailwindCSS & ShadCN UI Categorized project filtering (Web, AI/ML, Blockchain, Data Science) Hover-based stack previews for better UX Portfolio-ready design for developers

CodeCraft – Interactive Coding Platform

CodeCraft is a full-stack, browser-based online IDE inspired by Visual Studio Code that allows users to write, execute, and share code snippets securely in real-time with authentication, collaboration features, and SaaS-style monetization. Key Contributions • Built a real-time cloud-based IDE platform using Monaco Editor with multi-language support and theme customization • Implemented secure code execution pipeline using Piston API with isolated Docker containers for sandboxed runtime execution • Integrated real-time database architecture using Convex for instant snippet syncing, comments, stars, and analytics • Designed authentication system with Clerk supporting Google, GitHub, and email-based login • Developed subscription & payment flow using Lemon Squeezy, including webhook handling for premium feature access • Implemented state management using Zustand for editor preferences, themes, and execution state • Created responsive SaaS UI with Tailwind CSS + Framer Motion animations • Deployed scalable production system on Vercel with CI/CD pipeline Tech Stack Next.js 15, TypeScript, Convex, Clerk Auth, Monaco Editor, Tailwind CSS, Zustand, Framer Motion, Piston API, Lemon Squeezy, Vercel Impact • Enables real-time collaborative coding without local setup • Demonstrates SaaS monetization and cloud deployment pipeline • Showcases enterprise-level system design and full-stack engineering skills

CodeCraft — Online IDE Platform

Developed an end-to-end AI-powered system for lung cancer detection and classification using histopathology biopsy images, leveraging deep learning and explainable AI techniques. 🧠 Built and compared three models (Custom CNN, VGG16, ResNet50) using transfer learning, achieving up to ~98% accuracy with ResNet50. 🔬 Designed the system to classify images into: • Lung Adenocarcinoma (Malignant) • Squamous Cell Carcinoma (Malignant) • Normal Lung Tissue (Benign) 🔥 Integrated Grad-CAM (Explainable AI) to visualize model attention, enabling interpretability of predictions for medical insights. 📊 Implemented complete evaluation pipeline: • Confusion Matrix, ROC Curves, Precision/Recall/F1 • Balanced dataset (15,000 images – LC25000) • Data augmentation & optimized tf.data pipeline 🖥️ Developed a production-ready Streamlit web app with: • Real-time image upload & multi-model predictions • Confidence scores & probability visualization • Grad-CAM heatmaps for diagnosis explanation • Batch analysis & downloadable reports ⚙️ Tech Stack: Python, TensorFlow/Keras, OpenCV, Streamlit, NumPy, Pandas, Matplotlib, Seaborn 💡 Impact: This project demonstrates how AI can assist medical professionals by improving diagnostic efficiency, accuracy, and interpretability in cancer detection. ⚠️ Note: Built for research & educational purposes (not for clinical use).

Lung Cancer Detection & Classification using 

PrepWise is an AI-driven mock interview system that helps job seekers practice interviews through real-time voice interaction, personalized AI-generated questions, and detailed performance feedback. Key Features • Built secure user authentication using Firebase Auth & Firestore • Designed customizable AI interview creation (role, tech stack, difficulty, interview type) • Integrated real-time voice interview agent using Vapi API • Implemented Gemini AI for dynamic question generation and performance feedback • Developed automated AI feedback reports with score breakdowns, strengths, weaknesses, and improvement tips • Built responsive and modern UI using Tailwind CSS + Shadcn UI • Deployed production-ready application on Vercel with secure environment configurations Tech Stack Next.js 14 (App Router), React, Tailwind CSS, Shadcn UI, Firebase, Gemini AI, Vapi Voice API, Zod, React Hook Form, Vercel Impact • Enables realistic interview simulations using voice-based AI agents • Provides actionable feedback to improve communication, technical skills, and confidence • Helps bridge the gap between interview preparation and real-world performance

PrepWise — AI Mock Interview Platform

Developed a modern, scalable full-stack project management platform designed to streamline team collaboration, task tracking, and workflow management in a unified interface. This application enables teams to efficiently manage projects, assign tasks, track progress, and collaborate in real-time with advanced analytics and automation features. 🔍 Key Highlights: • Built a multi-workspace system supporting organizations and team-based collaboration • Designed project & task pipelines with status tracking (Todo → In Progress → Review → Done) • Implemented role-based access control (RBAC) for secure team management • Developed real-time collaboration features including task comments and activity tracking • Integrated email notifications for task assignments and deadline reminders using background jobs • Created interactive dashboards & analytics with charts and calendar views for productivity insights • Enabled Google OAuth authentication and secure session management • Built a responsive UI with light/dark mode for enhanced user experience • Deployed using Vercel with serverless backend architecture

🚀 ProJect — Full-Stack Project Management Ap

RESUMIND is a serverless, AI-powered resume analyzer that provides actionable feedback against job descriptions. It converts uploaded resumes into images, evaluates them with AI models, and generates insights on tone, style, completeness, skill alignment, and ATS (Applicant Tracking System) score. Key Features: AI Resume Analysis with Claude, GPT, and Gemini PDF upload & preview with drag-and-drop functionality Multi-resume management with ATS scoring Collapsible feedback sections for tone, style, skills, and structure Serverless & cost-free deployment powered by Puter.js

RESUMIND – AI-Powered Resume Analyzer

RESUMIND is a fully serverless, AI-powered resume analysis platform that evaluates resumes against job descriptions and provides ATS scoring, skill matching, and actionable feedback using modern cloud-native architecture. Key Features • Built serverless architecture using Puter.js (cloud storage, database, authentication & AI services) • Implemented AI-based resume evaluation system using Claude, GPT, and Gemini models • Developed drag-and-drop PDF upload system with automatic PDF-to-image conversion for visual preview • Designed ATS scoring engine to analyze resume structure, tone, skills alignment, and job relevance • Created multi-resume management dashboard with organized resume tracking and feedback display • Built collapsible feedback UI for structured recommendations (tone, content, skills, formatting) • Integrated secure authentication using Puter OAuth • Deployed production-ready serverless app on Vercel Tech Stack React.js, TypeScript, Tailwind CSS, Zustand, Puter.js (Serverless AI + Storage + Auth), Vercel Impact • Helps job seekers optimize resumes for ATS systems • Demonstrates serverless cloud architecture without backend infrastructure • Showcases AI integration in real-world SaaS-style applications

RESUMIND — AI-Powered Serverless Resume Analy

Spot is a full-stack AI-powered SaaS platform designed to simplify event creation, discovery, and management. It enables users to generate complete event details from a single prompt using Google Gemini AI, while providing a seamless experience for both organizers and attendees. 🚀 Key Contributions: • Built an AI-powered event generation system using Google Gemini API to auto-create event details (title, description, pricing, capacity) from a single prompt • Developed a real-time event management platform using Next.js and Convex with live dashboards for registrations, revenue, and attendee tracking • Implemented QR code-based ticketing & check-in system enabling seamless event entry with real-time validation • Designed a personalized event discovery engine based on user interests and location with search and filtering features • Integrated secure authentication and subscription system using Clerk with free and Pro tier access control • Deployed a scalable SaaS application on Vercel with serverless APIs and CI/CD pipeline

Spott  AI-Powered Event Management Platform

🚀 StegoShield is a Python-based image steganography tool that enables secure communication by embedding hidden messages inside images using the Least Significant Bit (LSB) technique. The tool ensures that messages remain undetectable while allowing password-protected retrieval. 💡 Key Features: ✔️ Stealthy communication with invisible message embedding ✔️ User-friendly GUI built with Tkinter ✔️ Password-protected decoding for added security ✔️ Supports multiple image formats (PNG, JPG, BMP) ✔️ Fast & efficient encoding & decoding process 🔹 Technology Used: Python, Tkinter, OpenCV, NumPy, LSB Steganography 🔹 Future Scope: AES encryption, audio & video steganography, mobile app integration

StegoShield – A Secure Image Steganography To

WebSmith AI is an intelligent website builder that empowers users to create and customize professional websites instantly using natural language prompts. Built with Next.js, React, Tailwind CSS, and OpenAI API, it leverages AI to generate responsive designs, manage content dynamically, and deploy sites seamlessly to Vercel. The project demonstrates full-stack integration, scalable architecture, and modern UI/UX design for effortless web creation. 🔗 Live Demo: https://web-smith-ai.vercel.app/ 💻 GitHub Repository: https://github.com/Saket22-CS/WebSmithAI

WebSmith AI – AI-Powered Website Builder

WebSmithAI is a full-stack AI-powered SaaS platform that enables users to generate, edit, and deploy production-ready websites in real time using natural language prompts and inline visual editing. Key Contributions • Built a real-time AI website generation system using OpenRouter streaming APIs to generate responsive Tailwind-based layouts instantly • Designed a unique inline editing engine allowing direct modification of website elements (text, spacing, colors, fonts) without re-triggering AI generation • Implemented secure SaaS authentication and subscription billing using Clerk with credit-based usage limits • Engineered a scalable backend using Drizzle ORM + Neon PostgreSQL to persist projects, design frames, and AI chat history • Integrated AI-powered image processing pipeline using ImageKit for background removal, upscaling, and dynamic transformations • Deployed production system on Vercel with CI/CD, environment security, and serverless API architecture Tech Stack Next.js 14, TypeScript, Tailwind CSS, Clerk Auth, Drizzle ORM, Neon PostgreSQL, OpenRouter AI, ImageKit, Vercel Impact • Reduced website creation time from hours to seconds • Enabled non-technical users to build and customize responsive websites • Demonstrated end-to-end SaaS architecture with AI + cloud deployment

WebSmithAI — AI SaaS Website Generator

YOOM is a production-ready video conferencing web application built using Next.js 14, Stream Video SDK, and Clerk Authentication, offering real-time meetings, scheduling, recordings, and participant management with a modern UI. Key Features • Implemented secure authentication system using Clerk (Google, GitHub, Email login) • Built real-time video conferencing with Stream Video SDK • Developed instant meeting creation and scheduled meeting booking system • Added meeting recording access and history tracking • Created personal permanent meeting rooms with shareable links • Integrated advanced in-meeting controls (mic/camera toggle, screen share, emoji reactions, participant pinning) • Designed dynamic meeting layouts (Grid, Speaker Left/Right views) • Built fully responsive UI using Tailwind CSS + ShadCN UI • Deployed scalable production build on Vercel Tech Stack Next.js 14, TypeScript, Stream Video SDK, Clerk Auth, Tailwind CSS, ShadCN UI, Vercel Impact • Enables real-time video collaboration similar to Zoom • Demonstrates enterprise-grade authentication and live communication systems • Showcases scalable frontend architecture using modern React frameworks

YOOM — Enterprise-Ready Video Conferencing Pl

YOOM is a modern real-time video conferencing application inspired by Zoom, built using Next.js 14, TypeScript, and Tailwind CSS. 🚀 Key Features: 🔐 Authentication & User Management with Clerk 🎥 High-quality video conferencing powered by Stream 💬 Real-time chat inside meetings 🗂️ Upcoming & Previous meetings dashboard 📼 Meeting recording & playback support 👤 Personal meeting rooms for one-click collaboration 🌐 Fully responsive and scalable with serverless architecture This project demonstrates expertise in full-stack web development, real-time communication, and modern cloud integrations.