AI Engineering
Senior AI Engineer
Build AI automations that run in production for real clients — owning the design, the evaluation harness and the operational reality, not just the prototype.
- Location
- Bengaluru
- Arrangement
- hybridfull-time
- Level
- Senior (5+ years)
- Compensation
- ₹38–58 LPA, plus performance bonus
- Posted
- 15 July 2026
Why this role exists
Most AI engineering roles stop at the prototype. Someone else worries about accuracy in production, cost per user, what happens when the provider is down, and who fixes it at 3am.
Here you own all of that. It is a harder job and a considerably more interesting one, because you find out whether what you built actually worked.
What you will actually spend time on
Roughly a third on client discovery and design — sitting with the people doing the work, finding out how the process really runs. A third building. A third on the operational reality: evaluation, cost, observability, handover.
If you want to spend all your time on model experimentation, this is the wrong role and we would rather you know that now.
What we look for
Engineers who are sceptical of their own systems. The strongest signal in an interview is someone who describes a project they shipped and can immediately tell us its accuracy, its cost per request, and the two ways it fails.
We also value the willingness to argue that a project should not happen. Roughly a third of our audits conclude AI is not the answer, and that only works if the engineers saying so are respected for it.
Process
Application review within a week. A short paid take-home based on real work — not a puzzle. Then two conversations: one technical, one about how you like to work. Four steps, usually inside three weeks.
What you will do
- Design and ship AI automations end to end, from process discovery to production
- Build evaluation harnesses before building features, and defend the numbers
- Integrate with client systems that were not designed to be integrated with
- Own cost and latency budgets for the systems you build
- Write the runbook and train the client team who inherits your work
- Tell a client honestly when AI is the wrong answer for their problem
What we are looking for
- Five or more years building production software, some of it with LLMs
- Strong Python or TypeScript, and comfort in both is a real advantage
- Experience with retrieval systems and why naive vector search stops working
- Able to reason about cost per request, not just whether it works
- Direct client communication — this is not a role behind a product manager
- Willing to say a project should not go ahead
Apply
No CV upload. Tell us what you have built and we will read it.