Headline

Full-Stack Engineer building AI Agents & Products | LLMs · RAG · TypeScript · Python

About

𝘐 𝘭𝘪𝘬𝘦 𝘣𝘶𝘪𝘭𝘥𝘪𝘯𝘨 𝘵𝘩𝘪𝘯𝘨𝘴 𝘵𝘩𝘢𝘵 𝘴𝘪𝘵 𝘴𝘰𝘮𝘦𝘸𝘩𝘦𝘳𝘦 𝘣𝘦𝘵𝘸𝘦𝘦𝘯 𝘈𝘐, 𝘴𝘰𝘧𝘵𝘸𝘢𝘳𝘦 𝘢𝘯𝘥 𝘳𝘦𝘢𝘭-𝘸𝘰𝘳𝘭𝘥 𝘱𝘳𝘰𝘣𝘭𝘦𝘮𝘴. 𝗪𝗵𝗮𝘁 𝗜 𝗵𝗮𝘃𝗲 𝗯𝘂𝗶𝗹𝘁 • Cuemath's games service for 𝟱𝟬𝗞+ 𝘂𝘀𝗲𝗿𝘀, 𝟱𝗞+ 𝗱𝗮𝗶𝗹𝘆 𝗮𝗰𝘁𝗶𝘃𝗲 across web and mobile, using React, TypeScript and Phaser on the frontend, Node.js (Colyseus) for game servers, and FastAPI for other backend services. • Automation for detection of teacher-session issues including camera framing, teacher presence, lighting, closed eyes with OpenCV and MediaPipe. Reached an 𝗙𝟭 𝗼𝗳 𝟬.𝟵𝟬 and packaged as a reuseable module. • Bunch of things with LLMs that are much more interesting than another chatbot. 𝗖𝘂𝗿𝗿𝗲𝗻𝘁𝗹𝘆 These days I'm building with agents, RAG, tool calling, computer vision and full-stack systems — usually with TypeScript/Python on one end and a working product on the other.

Location

Bengaluru

AI stack

claude code

claude

github copilot

langgraph

ChatGPT

Ragas

Work experience

SDE at Cuemath, Jun 2024 - Apr 2026 · 1 yr 11 mos, On-site, • Owned the games service for Cuemath’s learning platform (𝟱𝟬𝗞+ 𝘂𝘀𝗲𝗿𝘀, 𝟱𝗞+ 𝗱𝗮𝗶𝗹𝘆 𝗮𝗰𝘁𝗶𝘃𝗲) across web and mobile, using React, TypeScript and Phaser on the frontend, Node.js (Colyseus) for game servers, and FastAPI for other backend services. • Automated detection of teacher-session issues including camera framing, teacher presence, lighting, closed eyes with OpenCV and MediaPipe. Reached an 𝗙𝟭 𝗼𝗳 𝟬.𝟵𝟬 (𝟵𝟯% 𝗽𝗿𝗲𝗰𝗶𝘀𝗶𝗼𝗻, 𝟴𝟳% 𝗿𝗲𝗰𝗮𝗹𝗹) in testing on past session recordings and a gated pilot with 50+ teachers, then packaged it as a reusable internal library. • Designed a base game framework that handles event dispatching, state updates, animations, turn changes, and real-time sync, so a new game comes down to roughly 𝟳𝟱% 𝗹𝗲𝘀𝘀 𝗲𝗳𝗳𝗼𝗿𝘁 per game. • Built the real-time layer for multiplayer games, first on Colyseus (Node.js, WebSockets), then migrated it to PubNub(pub/sub): shared game state, scoring, timers, and leaderboards.

SDE - Intern at Cuemath, Nov 2023 - Jun 2024 · 8 mos, Hybrid, • Shipped 𝟭𝟳+ 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗴𝗮𝗺𝗲𝘀, including multiplayer, puzzle, and math-based experiences using React and Phaser. • Implemented WebSocket-based real-time gameplay for multiplayer games (live updates, synchronized state) and created reusable utilities and modular components shared across games.

Web-ML Intern at FrugalX, May 2023 - Jul 2023 · 3 mos, Banglore, • Built a 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝟭:𝟭 𝘃𝗶𝗱𝗲𝗼 𝗰𝗼𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗶𝗻𝗴 𝗮𝗽𝗽 with WebRTC and PeerJS, combining low-latency audio/video, computer-vision-driven input, and 3D avatar interactions.

Artificial Intelligence Trainee at Forsk Coding School, Jul 2022 - Aug 2022 · 2 mos, Jaipur, Rajasthan, India, • Applied NLP and ML techniques to analyze customer feedback and sentiment at Forsk. By visualizing data using DASH and deploying solutions on AWS, for informed decision-making.

Education

B.E. / B.Tech, Computational Science, rajasthan technical university

Public links

github: https://github.com/mrobox7

leetcode: https://leetcode.com/mrobox7

mohitbajaj.vercel.app: https://mohitbajaj.vercel.app/

medium: https://medium.com/@mrobox7