Staff Software Engineer

About the Company A Series A applied AI company in London, building voice agents that answer a business's phone, hold the conversation and complete the work behind it, without a person on the line. The callers are customers of service businesses who still pick up the phone. The company is under twenty people, with a small engineering team, several engineering hires planned, and a product that is live with customers. The Role Voice agents that handle the call end to end Answer the call, hold the conversation, complete the work behind it, and update the business's systems, without a person on the line. All of it in real time, while the caller waits. This is a Staff role on a small engineering team, hired to build features across the whole stack and to raise the standard of how the team builds. The person works directly with customers to understand what they need, scopes the solution, ships it, and owns it in production. The judgement the role turns on is when to ship a rough first version and when to build the foundation. This person makes that call explicitly, feature by feature, and is accountable for how it turns out. What is actually unsolved A voice agent acts in the real world while the caller is still on the line. A model call that times out after the provider has done the work, a booking written twice, an update that half completed: these are the product's daily failure modes, and each one is heard by a real person on the phone. The unsolved problem is making an autonomous agent's actions correct and recoverable under those conditions, across model, speech and telephony providers that fail independently, on a team that is shipping and growing at the same time. Responsibilities Build and ship core product features across the stack: Python services, a TypeScript front end, cloud infrastructure, event-driven systems and third-party integrations.Own each feature from the first customer conversation to production, and stay accountable once it is live.Decide, per piece of work, whether a rough first version or a durable foundation is the right build, and make that call explicit to the team.Make the calls to model, speech and telephony providers safe under failure: timeouts, retries, and side effects that must not happen twice.Build the integrations with the systems customers already run on, and keep them correct when the other side fails.Hold the bar on testing, documentation, reliability and security, and bring the rest of the team to it. Requirements This is a role for product engineers with a startup background: people who built a product they owned and ran it in production. Consultancy and software-house experience, delivering for clients, is not the background it is built on.You have shipped an LLM-based product to production end-to-end and run it with real users, from the model calls and the orchestration around them to the fixes after launch. Agentic systems, not isolated model work.Async Python is your production language: asyncio services, httpx, provider SDKs such as AsyncOpenAI, and error handling that survives timeouts and partial failures against non-deterministic responses.A TypeScript front end at working fluency, and comfort owning a feature on both sides.You have sat with non-technical customers, worked out what they needed, scoped the solution and delivered it.5 to 10 years of full-stack engineering, backend-leaning, including AI product work.You have worked at Series A or earlier, owning outcomes rather than tickets.Production systems on Kubernetes, serverless, and distributed infrastructure. Nice to Have Speech-to-text, text-to-speech, telephony or event-driven systems in production.Integrations with CRM, scheduling or billing systems, built with the customer in the room.Founder or early-employee experience.A computer science degree. Culture The team works together in a London office five days a week, 9 to 6. That is the working pattern, not a preference, and it suits people who want to be in the room while the product is built and who choose that over remote flexibility. The pace is that of a company that has just raised and is hiring across engineering. It is demanding, and the team says so. In return: a small team, direct access to customers, and ownership of what you ship. Apply or contact me directly