Engineering
Software Development Intern
Ship real features across a Python + React product used by students and teachers in Indian schools every day — with a founder reviewing your code.
About the role
eSkolr is an AI platform for schools. Its headline is a tutor that answers a student’s doubt only from their own textbook — but the same understanding of the books also marks homework (including handwritten photos), sets question papers in the board pattern, builds flashcards and mind maps, plans each student’s week and drafts report-card remarks. It is live with schools, so the code you write is used by real students and teachers the week it ships.
You will work across the stack — a Python (FastAPI, PostgreSQL) backend, a React web app and a React Native mobile app — on a small team where every pull request is reviewed and every feature is verified against real textbooks before release.
What you will do
- Build and ship features across the backend (Python, FastAPI, SQLAlchemy, PostgreSQL) and the web / mobile clients (React, React Native, TypeScript).
- Write tests for what you build and fix what you break — we care about correctness more than speed.
- Help improve how the product reads textbooks and answers questions: chunking, retrieval and answer quality, measured against real chapters.
- Take part in daily stand-ups, code reviews and planning; write clear notes so the next person can follow your work.
- Follow our security and data-handling practices — the product holds data about minors and that is a responsibility, not a formality.
What we are looking for
- A student or recent graduate in Computer Science / IT or an equivalent background, available full-time (8 hours a day, Monday to Friday, IST) for 6 months.
- Solid fundamentals in Python or JavaScript / TypeScript, and comfort with Git and the command line.
- You have built something end to end — a project, a hackathon entry, an open-source contribution — and can walk us through it.
- Clear written English; you ask when something is unclear rather than guessing.
- A reliable computer and internet connection (the internship is fully remote).
Nice to have
None of these is required. If you have them, tell us.
- Exposure to FastAPI, React or React Native.
- Any experience with LLM APIs, embeddings or retrieval-augmented generation.
- Familiarity with PostgreSQL, Redis, Docker or CI pipelines.
- You studied from NCERT / state-board textbooks and remember what being stuck at 10 pm felt like.
Problems you would work on
Real items from our board — not interview exercises.
Many Hindi textbooks are typeset in legacy fonts, so the text you extract is gibberish. Turning that into clean Unicode the search can understand — and keeping citations lined up with the printed page — is a real, unglamorous, essential problem.
A student photographs their notebook; we read the handwriting, mark the answer against the scheme, flag the words we were unsure of and send borderline cases to the teacher. Every piece of that can go wrong in an interesting way.
Semantic search cannot see numbers — “10.1” and “10.9” look identical to it. Resolving a numbered reference to the exact exercise, and honestly saying “not in this book” when it isn’t, is the difference between a tutor and a guesser.
Sections, marks, question types, a printable A4 paper and a separate answer key with a marking scheme — generated from the chapter, laid out so a school can photocopy it on Monday.
Retrieval, reranking and generation all compete for the same hardware. Caching that never serves one student another’s answer, streaming, and knowing exactly where the seconds go.
Multi-school isolation is enforced with row-level security in PostgreSQL, below the application, so a bug in app code cannot leak a school’s data. Every new table and worker has to respect it.
When a Maths question needs a construction or a graph, the answer should come with one — computed exactly, drawn on the client, and refused outright rather than drawn wrong.
A student asks in Roman-script Hindi about an English chapter. Retrieve in the book’s language, answer in the student’s, keep the citation, and score confidence fairly across scripts.
The stack
Compensation
₹8,000 per month stipend · internship certificate and a letter of recommendation on completion · full-time offer for interns who do well
Every term is set out in a written offer / engagement letter before you start. Applications accepted until 31 January 2027.
How we hire
- 1Apply by emailSend the items listed on the role page. We reply to every application, usually within five working days.
- 2A 30-minute introduction callWe tell you honestly what the role and the company are like; you tell us about your work. Both sides decide whether to continue.
- 3A short practical exerciseEngineering: a small take-home task (two to four hours, on something close to our real work). Partnerships: a short role-play of a call with a principal.
- 4Conversation with the founder, then an offerA final conversation about expectations on both sides. If we both say yes, you receive a written offer letter with every term spelled out.
eSkolr is an equal-opportunity employer. We do not discriminate on the basis of gender, religion, caste, region, language, age, disability, marital status or any other characteristic protected by law. Your application data is used only to assess your application and is kept for up to 12 months; email us at any time to have it deleted.