LLM / AI Engineer

Our voice AI system is live in production - real calls, real borrowers, Indian languages. We're past 0→1. Now we're scaling it, hardening it, and building the next generation of AI-powered debt collection.

This is an operator role, not a research role. You'll work on a system already handling live calls. Your job: make it smarter, cheaper, and more resilient.

What you'll own → LLM integration, prompt design, and negotiation workflows. Per-call token cost optimisation across model tiers, transcript analysis & failure pattern identification. Sentiment-based workflow tuning - tone shifts, resistance cues, drop-offs. You'll own the outcomes, not just the outputs - resolution rate, PTP conversion, handoff precision, and the signal buried in every call that explains why those numbers move.

What we're looking for → 2–5 years engineering, 1–2 years with LLM APIs in production, prompt engineering, structured outputs, tool/function calling, hands-on Python for scripting, API work, and data analysis.

Why this role You won't be building in a vacuum. The system is live, the problems are real, and you'll own the AI layer end-to-end - shaping scaling decisions and defining what comes next. No notebooks, no demos. Just production LLMs.

We're building this team from scratch - if you know someone who'd be a fit, we'd love to hear from them too.