Senior Software Engineer, Agents

About the role

At Replicant, we believe AI should work for people, starting with customer service. That's why we built a platform that helps contact centers resolve more requests, proactively identify issues, and improve agent performance with AI-powered conversation intelligence and AI agents that act like your best reps.

Our AI agents handle millions of calls every month for Fortune 500 companies and high-growth innovators. From processing payments to booking appointments and authenticating users, they help customers get what they need instantly, 24/7. Meanwhile, our real-time conversation insights help contact center leaders coach better and improve every interaction.

We are leading the shift from legacy systems to AI-first service, powered by large language models (LLMs) and designed for enterprise scale, security, and empathy. If you're excited by the potential of LLMs, voice AI, and building category-defining technology with a kind, ambitious team, you'll love it here.

Our customers don't buy software. They buy a working AI agent that answers their calls, resolves their customers' problems, and holds up in production. As a Senior Software Engineer, Agents on our Delivery team, you make that happen for specific customers: you sit with them to understand their workflows, design and write the agent's instructions and conversation behavior, build the integrations the agent calls as tools, and own the agent end to end from blank prompt to production. This is a customer-embedded engineering role, equal parts trusted advisor, integration builder, and prompt engineer, and you are measured on customer outcomes, not features shipped. It's also one of the most in-demand emerging skill sets in the market, and you'll build it on real production traffic, not demos, with real autonomy on a team small enough that your work is visible to customers and leadership alike.

What You'll Do

- Own AI agents for your assigned customers across the full lifecycle: discovery, design, build, evaluation, launch, and live operation.

- Sit with customers, map their workflows, and translate ambiguous business goals into a concrete agent scope: what the agent handles, what it escalates, and what success looks like. Uncover the customer's real business rules, including the unwritten ones, and pin them down so the agent enforces that business logic strictly.

- Design and write the agent's prompts, instructions, and conversation flows, and iterate against evals, real transcripts, and your own product taste for what a good conversation sounds like.

- Wire the agent into customer systems (CRMs, ticketing, internal services): call their APIs as tools, shape inputs and outputs so tool calls are reliable, and handle auth, errors, and edge cases. You won't build the customer's APIs, but you will tell them what the agent needs from an API, then test and validate what they deliver.

- Build test sets from real cases covering not just the happy path but unhappy paths, edge cases, and hallucination risks; measure agent quality and harden agents for production: guardrails, failure modes, latency, and cost.