engineer

Arrowhead is building AI voice agents for sales and support teams in India. The product sits at the intersection of conversational AI and enterprise workflows — a space moving fast, with real deployment complexity underneath the surface.

Arrowhead is a small, fast-moving team building AI voice agents for sales and support teams in India. The backend is the product — and right now, the team is short one engineer who can hold that weight.

This is a backfill hire, but the scope is real: you'll work directly on the AI layer, own backend systems that sit under live voice interactions, and move fast in a codebase where the margin for slow is thin. Series A, single-digit team, Delhi.

• The backend systems under Arrowhead's voice agents — the infrastructure that makes real-time AI conversations work at the application layer.
• LLM integration work — connecting language models to live systems, not prototyping in isolation.
• Code quality on your surface area — you hold the bar for what ships from your end.
• Debugging and incident response on the systems you own; the team is small, so ownership is real.
• Input into architecture decisions — at this team size, your opinion on how to build something matters and gets heard.

• You've built and shipped backend systems in Python, and you're comfortable across Java and Golang when the job calls for it.
• You've worked with LLMs in production — not just fine-tuned a model in a notebook, but integrated one into a system that real users hit.
• You move fast without leaving a mess behind — you know the difference between a shortcut and a trap.
• You're comfortable being one of a handful of engineers; you don't need a large team around you to do your best work.
• You read a broken system and know where to start — debugging is instinct, not process.
• You've worked in a startup or a small team before and you know what it means to own something end to end.
• You ask sharp questions before you write a line of code.

• Ship, then sharpen. Get working code out fast; iterate from there, not from a whiteboard.
• Own the problem, not just the ticket. If something is broken upstream of your task, you flag it or fix it.
• Write code others can read. The team is small; nobody has time to decode clever.
• Communicate in context. Short, specific updates beat long status reports.
• Bring a point of view. When you see a better way, say so — with a reason.
• Stay close to the product. Understanding what the voice agent is doing for the end user makes your backend decisions sharper.

Competitive cash at ₹11–12L per year, with the equity and scope that come with a Series A team in single digits. You'll be one of a handful of engineers building the backend of a live AI voice product.