THE OPPORTUNITY

The way most software teams are organised is a relic. The sprint cadences, the ticket grooming, the review bottlenecks — all built for a world where individual output was low and coordination was the hard problem. AI inverted that. The hard problem now is architectural judgment at high velocity, and most organisations have no idea what to do with the engineers who have it. If you are one of those engineers, this posting is for you.

Sujho is an AI Teaching Assistant platform for Indian K-12 education. We give tuition teachers an army of AI-powered assistants for tutoring, grading, lesson planning, parent communication, and student analytics. Three AI agents serve teachers, students, and parents through a shared data layer that compounds with every classroom interaction. The product is live on WhatsApp, serving CBSE Class 6–12. We are hiring our founding engineer to build the next layer.

THE ROLE

You will work directly alongside the founder — a software engineer who built the entire current stack. He owns product direction. You own the full-stack engineering underneath: APIs, data layer, web surfaces, infrastructure, and the agent harnesses everything runs on.

This is an AI-native engineering role. You will use Cursor, Codex, Claude Code and whatever comes next as core tools for every task. The expectation is that you ship at a velocity that would have been impossible a year ago — and that you do it with the architectural taste to keep the codebase clean as it scales. That combination is the core requirement for this role.

WHAT YOU’LL ACTUALLY DO

The core engineering challenge at Sujho is building a data substrate that gets smarter with every classroom interaction. Every assignment a teacher creates, every doubt a student asks, every score the system grades — all of it writes to a structured layer that the AI agents read from. The product teaches and tests simultaneously, which means every interaction is both an output to a user and a signal back into the system. Your job is to make that loop compound.

In practice, this means: designing data models that serve three different stakeholders through one shared substrate. Building agent infrastructure that is reliable, stateful, and improves as the data underneath it grows. Creating web surfaces that make a rich, interconnected dataset navigable and useful. Architecting systems for proactivity, where the product reaches out to users based on patterns it detects, rather than waiting for them to ask.

The specific features will evolve. The patterns underneath — a compounding data layer, recursive self-improvement through usage, multi-stakeholder agent architecture — are durable. You are building the infrastructure for those patterns, and in a system that compounds, the quality of early architectural decisions determines everything that comes after.

WHERE THIS GOES

Months 1–3: You are shipping alongside the founder, absorbing context, understanding how teachers and students use the product. By month three, you own the codebase end to end.

Months 4–12: You are running engineering. You are making the architectural calls that will outlast individual features. You are making your first hire. And you are designing from scratch what an AI-native engineering organisation looks like — the tooling, the conventions, the team shape — without legacy to retrofit.

Year 2+: The engineering team has built around you. Your focus is architecture, culture and strategy. The way this team ships will look alien to engineers at traditional companies — because you built it for a world they haven’t caught up to yet. We are taking on deep technical challenges that are unthinkable today.

The people that join companies at this stage and stick around tend to end up running things.

WHAT YOU NEED (DAY ONE)

• 2–5 years building production software. Systems that real users depend on.
• Strong full-stack fundamentals — Python backends, web frontends, databases, APIs, deployment.
• AI-native development workflow. Cursor or equivalent daily. Strong opinions on using it well.
• Experience with LLM-based systems — tool calling, structured outputs, agent harnesses.
• Architectural taste. Clean abstractions, thoughtful data models, code that reads well.
• The ability to take a product goal and figure out the engineering path yourself.
• Bonus: founding/early startup experience, agent-based systems, interest in K-12 education, public track record.
• Available in Gurgaon, in-person, 5 days a week.

THE DETAILS

Location

Gurgaon, Haryana (in-person)

Start Date

August 2026

Reports To

Founder & CEO directly

HOW WE HIRE

We want to see you build, not perform in interviews.

Step 1: A short conversation. 30 minutes with the founder. How you work, how you use AI tools, what you have built recently.

Step 2: A real engineering task. A hard take-home that mirrors actual Sujho engineering — an LLM integration, a data layer, an API, and a frontend. Use whatever tools you want. We evaluate architecture and judgment.

Step 3: A code review. Walk the founder through your assignment. What you built, what tradeoffs you made, what you would change, where AI got it wrong.

Step 4: A team conversation. Meet the rest of the team. Culture fit, communication style, how you work with non-engineers.

Pay: ₹800,000.00 - ₹2,000,000.00 per year

Work Location: In person