Projects

Strivo

Full-stack social media platform

  • Developing Strivo, a full-stack social media platform that integrates long-form video uploads and tweet-style posts within a unified system.
  • Designing and implementing scalable backend APIs to manage users, posts, media uploads, and interactions.
  • Building a modern, responsive frontend using React with advanced state and data-fetching patterns.
  • Implementing efficient client-server data synchronization using TanStack Query for optimized performance and caching.
  • Structuring the project with modular architecture to support scalability and future feature expansion.
React.jsTanStack QueryNode.jsExpress.jsCloudinaryMongoDBGit & GitHubMulterREST APIsJavaScriptTailwind CSS

PrepVector

AI-powered mock interview platform

  • Building PrepVector, a full-stack AI platform that unifies resume building with realistic, voice-driven mock interview simulation.
  • Designing AI pipelines using LangChain to generate tailored interview questions and evaluate spoken answers in real time.
  • Implementing a voice-driven interview simulator using native Web Speech APIs for text-to-speech and speech-to-text interaction.
  • Developing structured AI output parsing with Zod to reliably generate scorecards, skill-gap analyses, and multi-day preparation roadmaps.
  • Building an HTML-to-PDF resume generation pipeline using headless Puppeteer for pixel-perfect, ATS-friendly exports.
  • Architecting a cross-domain authentication system to ensure reliable session handling across separately hosted frontend and backend services.
  • Structuring the project with a scalable, modular backend and a feature-based frontend architecture for maintainability and growth.
React.jsTanStack QueryNode.jsExpress.jsMongoDBLangChainZustandPuppeteerTypeScriptTailwind CSSJWT AuthenticationREST APIs

FormForge

Schema-driven type-safe form builder

  • Building FormForge, a schema-driven form-building platform designed for full end-to-end type safety across the stack.
  • Designing a type-safe API layer using tRPC to eliminate contract drift between client and server.
  • Implementing dynamic form generation and validation using Zod schemas, enabling forms to be built and versioned like code.
  • Structuring the project as a Turborepo monorepo with modular, reusable packages for UI, database, and validation logic.
  • Integrating Drizzle ORM with Supabase for type-safe database schema management and authentication.
  • Building GitHub-integrated schema versioning to track and manage form schema history directly through repositories.
  • Developing a modern, responsive UI using shadcn/ui and Framer Motion for smooth, polished interactions.
Next.jsTypeScripttRPCDrizzle ORMSupabaseZodTurborepoTailwind CSSshadcn/uiFramer Motion

MultiAgent AI Research

Autonomous multi-agent research pipeline

Coming Soon
  • Building a multi-agent AI research pipeline that autonomously searches, scrapes, synthesizes, and self-critiques detailed research reports.
  • Designing a sequential 4-agent architectureSearch Scout, Reader, Writer, and Critic — where each agent owns a distinct stage of the research workflow.
  • Integrating the Tavily Search API to retrieve high-quality, relevant sources for each research query.
  • Building a web scraping pipeline using Cheerio to clean and extract readable content from raw HTML.
  • Implementing a LangChain-powered synthesis layer to compile scraped research into structured, cite-able markdown reports.
  • Developing a self-critique scoring system where a dedicated Critic agent grades each report, highlighting strengths, improvements, and a final verdict.
  • Adding security guardrails including prompt-injection detection and rate limiting to prevent API abuse.
  • Designing a premium animated dark-mode interface with word-by-word streaming report rendering using Framer Motion.
React.jsTypeScriptNode.jsExpress.jsLangChainTavily APICheerioFramer MotionTailwind CSSREST APIs

TwitterClone

Full-stack microblogging platform

  • Building TwitterClone, a full-stack microblogging platform replicating core Twitter/X mechanics with real-time social feeds.
  • Designing scalable backend APIs to manage users, tweets, likes, bookmarks, and follower/following relationships.
  • Implementing two dynamic feed systems — a global discovery timeline and a personalized following feed.
  • Building a media upload pipeline integrating Multer and Cloudinary for avatar and tweet media handling.
  • Implementing optimistic UI updates for likes and bookmarks to deliver instant, snappy user interactions.
  • Managing global client state with Redux Toolkit and redux-persist to ensure sessions and cached data survive page refreshes.
  • Developing secure JWT-based authentication with HTTP-only cookies and protected route middleware.
React.jsRedux ToolkitNode.jsExpress.jsMongoDBCloudinaryJWT AuthenticationTailwind CSSREST APIs

Snake Game

Vanilla JS DOM-based snake game

  • Building a classic Snake game entirely in vanilla JavaScript, with no canvas, game engine, or external dependencies.
  • Designing a DOM-based rendering system using a coordinate-keyed hash map to cache grid cells and avoid repeated DOM queries during the render loop.
  • Implementing precise collision detection logic, including tail-aware self-collision handling for accurate gameplay physics.
  • Building a fail-safe food-spawning system to guarantee valid placement without risking infinite loops.
  • Developing a persistent high-score system using browser localStorage.
  • Styling a modern glassmorphism UI with custom CSS animations, gradients, and responsive Flexbox/Grid layouts.
JavaScriptHTML5CSS3DOM ManipulationBrowser LocalStorage

Kanban Board

Zero-dependency drag & drop task manager

  • Building a lightweight, zero-dependency Kanban task management app using vanilla HTML, CSS, and JavaScript.
  • Implementing native HTML5 Drag & Drop functionality to move tasks across To Do, In Progress, and Done columns.
  • Designing a DOM-as-source-of-truth architecture, where the DOM is updated first and application state is rebuilt by querying the live DOM.
  • Building a synchronous persistence layer using localStorage, ensuring task data survives page refreshes without a backend.
  • Implementing event delegation for efficient task deletion, avoiding per-element event listeners.
  • Structuring the project with clean, semantic HTML and CSS variables for a maintainable, framework-free codebase.
JavaScriptHTML5CSS3Drag & Drop APIDOM ManipulationBrowser LocalStorage

Contact Manager API

RESTful contacts API with per-user isolation

  • Building a RESTful backend API for managing personal contacts, with secure per-user data isolation.
  • Designing JWT-based authentication with bcrypt password hashing for secure user registration and login.
  • Implementing full CRUD operations for contacts, with ownership checks ensuring users can only modify their own data.
  • Structuring the codebase with a clean separation of controllers, middleware, and models for maintainability.
  • Developing centralized error handling middleware for consistent, structured JSON error responses across the API.
  • Designing protected routes using custom JWT verification middleware to secure sensitive endpoints.
Node.jsExpress.jsMongoDBMongooseJWT AuthenticationBcryptREST APIs

Restaurant API

Role-based food ordering backend API

  • Building a RESTful backend API for a food ordering platform, handling restaurants, menus, categories, and orders.
  • Designing role-based access control supporting client, admin, vendor, and driver user types.
  • Implementing full CRUD operations for restaurants, food categories, and menu items.
  • Building an order management system with cart-based order placement and admin-controlled status updates.
  • Developing secure JWT-based authentication with bcrypt password hashing and protected route middleware.
  • Structuring the codebase with a clean, modular separation of controllers, models, routes, and middleware.
Node.jsExpress.jsMongoDBMongooseJWT AuthenticationBcryptREST APIs

© 2026 Samarth Pagaria • All Rights Reserved