Python vs. C++ on Serverless
Benchmarking polyglot microservices on Cloud Run — a deep dive into real-time distributed systems.
Hey, I'm Mohit 👋
Forward Deployed AI Engineer at Supervity. I specialize in the "hard parts" of AI adoption — moving beyond the notebook to architect secure, scalable systems for Fortune 500s. The best compliment? "Your code actually works."
FastAPI + Next.js + PostgreSQL.
Containerized & production-ready.

I occupy a fun, chaotic space in the developer ecosystem: an undergrad who simultaneously operates as a Forward Deployed AI Engineer for enterprise clients. At Supervity, I get paid to argue with LLMs and make them work in production.
My philosophy is simple: Code > Slides. I don't believe in teaching theory unless I've deployed it. Whether I'm migrating a client from OpenAI to Gemini or helping someone debug a Cloud Run container at 2 AM during a hackathon, my goal is to bridge the gap between "academic theory" and "enterprise reality."
The best compliment? When a prototype I built for a demo becomes the production system that closes a $3M deal. That's happened more than once.
Fun fact: I authored the official walkthroughs for Labs 1-7 at Code Vipassana (ADK, MCP, A2A). I treat these tutorials like production code — they have to work.
You don't really understand something until you can teach it. That's why I do all three.
I build frameworks that fix the things that annoy me about AI. Multi-agent systems, real-time voice AI, RAG pipelines that actually work in production — not just in notebooks.
The 'Polite' AI Interviewer
My content exists to bridge the gap between official documentation and reality. I decode complex architectural concepts and turn them into actionable, step-by-step walkthroughs.
Settling the Speed Debate
I don't just mentor; I teach. And when things break at 2 AM during a hackathon, I'm 'The Unblocker' — the person who stays calm and finds the fix.
Technical Lead & Creator
Architecture decisions, production war stories, and the lessons that only come from shipping AI that real people actually use.
Benchmarking polyglot microservices on Cloud Run — a deep dive into real-time distributed systems.
How I combined Cloud Run WebSockets, Gemini 2.5, and hybrid video for sub-second conversational AI.
Why BigQuery Vector Search and Google ADK are the secret weapons for stateful, production-ready agents.
You don't truly understand Cloud Architecture until you can explain it to a room full of tired devs. There's a specific joy in taking a complex, bleeding-edge feature and breaking it down until a room full of developers nods in understanding.
During the Build N Blog Marathon, developers hit a critical blocking issue with GCP Quotas and IAM access for GPUs. I didn't wait for support — I took the stage, diagnosed the IAM bottleneck live, and guided the entire room through a resolution process. Everyone deployed their Cloud Run containers that day.