Rewas Khatri
Full-Stack Developer building clean, scalable web apps & AI
New to Contra
Rewas is ready for their next project!
Interactive Web-Based Tic-Tac-Toe | HTML5, CSS3, JavaScript (ES6)
Engineered an interactive, client-side two-player game featuring automated win-pattern validation across 8 distinct board vectors and tie-state handling.
Implemented efficient event-driven DOM manipulation and modular game-state tracking using pure vanilla JavaScript with zero external libraries.
Designed a clean, adaptive interface using modern CSS Grid/Flexbox to ensure consistent layout and touch responsiveness across mobile and desktop viewports.
0
6
CartOrchestrator is an autonomous e-commerce AI assistant built on a zero-trust financial architecture. Most commercial conversational bots fail at the transaction boundary—they hallucinate inventory, miscalculate discounts, or expose sensitive checkout logic. CartOrchestrator isolates the language model strictly to conversational orchestration while delegating all catalog lookups, cart mutations, discount logic, and payment settlements to tamper-proof server-side guardrails.
0
9
A real-time voice AI assistant powered by Gemma 4. Tinkr captures microphone audio in the browser using the Web Speech API (SpeechRecognition), sends the transcribed text to a Flask backend, streams back reasoning via Server-Sent Events (SSE), and plays back responses using browser Text-to-Speech (TTS). A local Python backend bridges authenticated Google services — Calendar, Tasks, and a SQLite database — so the AI can take real actions on your behalf.:: Youtube vedio link: https://www.youtube.com/watch?v=40pFkA7IlQ8&t=1s
0
12
KookHealth is an AI-powered recipe and nutrition app I built end-to-end — from backend architecture to a full brand and UI redesign. Users add the ingredients they already have and get a unique, AI-generated recipe (name, quantities, cook time, steps) via Google's Gemini API, or snap a photo of a meal to get an instant nutritional breakdown (calories, protein, carbs, fat, fiber, sugar).
I built the backend with Node.js/Express, keeping the Gemini API key fully server-side for security, and designed a two-theme (warm/earthy) visual system with a custom logo, bold typography, and scroll-based micro-interactions on the frontend. The app is deployed on Render with a CI-friendly deploy pipeline, and saved recipes/preferences persist locally per user.
This project reflects my ability to own a product from architecture to pixel-level design — integrating a real AI API into a production-ready, responsive, accessible web app.::
0
35