AI Full-Stack Engineer

We're building AI that actually talks to India — in Hindi, Marathi, Hinglish, and the languages people use at home, not just polished English. If you want your code to reach people who normally get left out of the tech party, keep reading.

How to apply

Please click on the link to apply - https://forms.gle/oKJHUe56F1XJBcA8A

The gist

You'll build AI products end-to-end — from the model calls and RAG pipelines to the screens people actually tap on. Small team, real users, no layers of approval between you and shipping. You write code in the morning, see it in someone's hands by next week.

If you've ever stayed up too late getting a side project to finally work — that's the energy we want.

What you'll actually build

●      AI assistants and tools that work in Indian languages — text and voice — over WhatsApp, Telegram, and the web

●      RAG and retrieval systems so our AI answers from real, trustworthy info instead of making things up

●      The full product: clean React/Next front-ends and solid Python APIs behind them

●      Pipelines that turn messy real-world data (audio, regional text, documents) into something useful

●      Whatever the problem needs — we care more about it working for the user than about doing it the "textbook" way

You won't be siloed into one corner of the stack. If you like seeing the whole thing, you'll like it here.

The stack

Here's what we work with day-to-day:

●      Frontend: React, Next.js, TypeScript, Tailwind CSS

●      Backend: Python, FastAPI, REST APIs (Flask/Django around the edges), OAuth, SQLAlchemy

●      AI / LLM: Claude API, OpenAI API, RAG, embeddings, vector search, prompt engineering

●      Databases: PostgreSQL, Supabase, pgvector, ChromaDB

●      Ship it: Git, GitHub, Docker, Vercel, Railway, GitHub Actions / CI-CD

●      How we move fast: Claude Code and Cursor are part of the workflow — we expect you to build with AI, not around it

You don't need all of this on day one (more on that below).

Who you are

The core stuff (the must-haves):

●      You can build a working app front-to-back — a React/Next UI and a Python (FastAPI) backend that talk to each other

●      You've actually used an LLM API in a project — calling Claude or GPT, wrangling prompts, maybe a basic RAG setup

●      You're comfortable with Git, databases (Postgres or similar), and getting something deployed

●      You can read a problem, make a sensible call, and ship — without waiting to be told every step

You don't need to check every box. If you've got the builder instinct and most of the core skills, the rest is learnable. We mean that — apply anyway.

Extra sauce (nice-to-haves)

Any of these and we'll be paying attention:

●      Indian-language AI: Hindi / Marathi / Hinglish NLP, AI4Bharat, IndicNLP, Sarvam — this is close to our heart

●      Voice: Whisper, ElevenLabs, speech-to-text and back

●      Chat platforms: WhatsApp Business API, Telegram Bot API

●      Agentic stuff: LangGraph, LlamaIndex, MCP

●      More tools: React Native / Flutter, Firebase, Redis, Celery, Sentry, pytest / Jest

What we care about beyond the code

This part matters as much as the tech:

●      You build things. Hackathons, GSoC, MLH / MLH Fellowship, open-source PRs, an indie project, something on Product Hunt — anything that shows you make stuff because you want to, not just because someone assigned it.

●      You care about who it's for. We work in social impact — NGO and grassroots-adjacent, vernacular, rural reach, accessibility. If "tech for people who usually get ignored" makes you sit up, you'll fit.

●      You're part of a dev community (or want to be). GDG, MLSA, GitHub Campus Expert, a college dev club, CodeChef / Codeforces — basically, you don't code alone in a cave.

●      You can explain your thinking. A tech blog (Medium, Hashnode, Substack), tutoring, mentoring, running a workshop — if you can teach an idea, you can work on a team.

●      You speak more than one language — literally. Hindi, Marathi, or any Indian language is a real plus when you're building for multilingual users.

What's in it for you

●      Ownership from day one — your name is on what ships

●      Real users, not a demo that lives in a drawer

●      Work with the newest AI tools instead of fighting legacy code

●      Mentorship + room to grow — we'll back you to get sharper, fast

●      [Add: comp band, work flexibility, leave, learning budget, equity, etc.]

We'd rather be upfront: this is small team energy. It's hands-on, it moves quick, and you'll wear a few hats. If that sounds like a feature, not a bug — we should talk.