About the team
Sarvam is building India's sovereign AI stack: large language models made for Indian languages and Indian contexts. The Education and Healthcare business unit takes that raw capability and turns it into products that teachers, students, doctors, and hospitals actually use. Think AI teaching assistants for government school systems, tutoring copilots, clinical documentation tools, and patient-facing assistants, all in the languages people actually speak.
This is where the model meets the messy real world, and you will be right in the middle of it.
What you will actually do
This is a full-stack role, which means you own features end to end -
Build product features across the whole stack: frontend interfaces, backend APIs, and the LLM logic that sits underneath
Ship AI-powered experiences such as RAG systems, agents, copilots, and document intelligence tools for education and healthcare use cases
Turn a rough idea from a pitch conversation into a working prototype
Write prompts, build evals, and actually measure whether the AI is doing a good job
Wire up integrations: internal tools, third-party APIs, and data pipelines
Hunt down bugs, chase weird edge cases, and speed up anything that feels slow or clunky
Sit close to the product and business teams, so you understand why you are building something, not just what
What We're Looking For
You do not need every single one of these. You do need to be genuinely comfortable in a few, and hungry to pick up the rest fast.
Core:
Strong Python (this is your bread and butter)
JavaScript or TypeScript, plus a modern frontend framework like React or Next.js
A backend framework like FastAPI, Flask, or similar
Comfort working with LLM APIs: prompting, function calling, structured outputs
SQL and general database sense (Postgres, ideally with some exposure to vector databases like pgvector, Pinecone, or Qdrant)
Git, and enough command line to not be scared of a terminal
Good to have:
RAG and agent frameworks (LangChain, LlamaIndex, or your own from-scratch setup, which we respect even more)
Docker and a working idea of how things get deployed
Cloud basics (AWS or similar)
Some ML depth: fine-tuning, PyTorch, model evaluation
Data tooling: pandas, and maybe a dashboard tool like Metabase
What makes you a fit
You can build. You have projects, side projects, hackathon wins, open source contributions, or something you can show us and talk about in real detail
You learn scary fast and are not thrown by a technology you have never seen before
Bonus: you have real opinions about AI products, and where they are genuinely useful versus hype
Why this is genuinely a great gig
You will learn more in a few months here than in a year almost anywhere else. You will work directly with people building at the frontier of Indian AI, ship things that matter, and watch your code touch education and healthcare at real scale. It is fast, it is occasionally chaotic, and it is a lot of fun. If you want a comfortable, slow, watch-from-the-sidelines internship, this is not it. If you want to build, learn, and be trusted with real responsibility from day one, this is it.