DevInterviewMasterLIVE
Tech interview prep platform: 10,000+ questions, sandboxed code execution, AI-generated video lessons, Razorpay subscriptions. Rust backend serving sub-10ms p99.
I'm Tanuj Mehta, an AI engineer with 5+ years shipping LLM-powered agents, RAG pipelines and full-stack platforms that real users pay for. Not demos, not notebooks: nine products live right now, with payments, auth, CDNs and monitoring, built and operated by me.
Every one of these is live right now, with real infrastructure behind it. Click any project for the full case study: architecture, hard problems, and what I learned running it in production.
Tech interview prep platform: 10,000+ questions, sandboxed code execution, AI-generated video lessons, Razorpay subscriptions. Rust backend serving sub-10ms p99.
System design learning platform with an AI mock interviewer that talks back: voice mode, drag-and-drop architecture canvas, 340+ HLD/LLD topics, in-browser code execution.
K-12 EdTech platform: adaptive quizzes, gamified arenas, and a zero-cost AI tutor running local LLMs with RAG over the syllabus. Java 21 backend with 186+ endpoints.
AI resume builder with an LLM ATS-scoring engine that gives section-level feedback, pixel-perfect Puppeteer PDF rendering, OAuth, and Razorpay payments.
Job automation pipeline: Playwright workers scrape boards, an LLM ranks role fit against your resume, and a published Chrome extension auto-fills applications.
Also live: superkinggames.com (real-time multiplayer Ludo over WebSockets), arthaaitech.com (automated stock-comparison video pipeline), and an HTML5 arcade game on CloudFront. Plus a self-hosted Jira clone, an autonomous engineering factory wired to it, a Postman clone, 5 Flutter/Flame games, voice-entry expense trackers, and a faceless Instagram reels factory on the Meta API.
I have been building with LLMs in production since 2024: at work, across my own products, and in automation pipelines that run daily without me.
Production ReAct-style agents with tool calling over a custom tool registry. Multi-step planning and execution at 90%+ task accuracy, shipped for enterprise at Kellton.
Deep hands-on with the Anthropic stack: Claude API tool use, Claude Code agents and skills, MCP servers, subagent orchestration, and prompt caching for cost control.
Retrieval over pgvector, ChromaDB and Pinecone. Built a fully local RAG tutor (Ollama + pgvector) that answers syllabus questions at zero API cost.
Voice interfaces that convert natural language into tool calls and DB queries across 50+ intents. AI interviewer with streaming TTS that starts speaking mid-response.
Evaluation harnesses, structured logging and observability for agent behavior. Offline-first design that cut cloud LLM costs by 60% while keeping 99.9% uptime.
Pipelines that generate at scale: 176 animated explainer videos (script to TTS to render), 10,000+ interview questions, and social content, all orchestrated by agent fleets.
I'm looking for remote AI engineering roles where I can own products end to end. If you want someone who treats your product like their own, let's talk.
tanujmehta2901@gmail.com