sequence - AI Engineer - London

• We care deeply about the quality of the product we're building. We're looking for engineers who are ambitious and take ownership end-to-end: from identifying what to build, through shipping, to making sure it works for real customers. • Here’s what you might work on: • Our AI infrastructure. We process over a billion dollars a year through Sequence so correctness is table-stakes. You'll build and own the platform that allows us to build reliable software on top of non-deterministic models, including agentic workflows, prompt iteration, tools, and evals. • AI-powered approval workflows. Enterprise deals need approval chains, but rigid workflows break when every company has different rules. You'll build flexible routing that customers configure in natural language - "require VP approval for deals over $100k with annual terms" - and turn that into deterministic, auditable business logic. • An intelligent collections agent. Chasing late payments is tedious, manual work. We're building an agent you instruct in natural language (“send a reminder at 7 days overdue; escalate at 14 days”), which then runs the entire workflow and collects the relevant context autonomously. • We're a lean team growing to 20+ engineers. You'll have real influence on what we build and how we build it. Early enough to shape fundamental architecture. Late enough that customers depend on what you ship. • We’re looking for product engineers who live in this space. You have informed views on which tool to reach for and when. You’ve built side projects to see what’s possible. You take AI security seriously - you’ve scrutinized the threat vectors and you treat models as untrusted by default. • See what we’ve shipped in our public changelog https://www.sequencehq.com/changelog. • YOU SHOULD APPLY IF • You're a builder who wants to solve problems end-to-end: • You've shipped LLM-based systems to real customers and have first-hand experience with how they fail in production • You've designed agentic systems and have informed views on the building blocks (like embeddings, memory, tool use, long-running state, recovery from partial failures) and when to reach for each • You’ve built evals infrastructure (like datasets, LLM as judge, prompt regression tests, monitoring) and have a clear view of what good looks like • You respect the weight of building business-critical systems on top of non-deterministic models and are comfortable writing resilient software • You have informed opinions about the tooling for building, evaluating, and securely running multi-provider AI systems in production • You've shipped production backend systems and care about the difference between "it works" and "it's reliable" • You care about customers - you want to understand why you're building something, not just what • You’re comfortable with ambiguity. You thrive in early ideation stages, share work-in-progress to gather feedback, and adapt easily based on input • You communicate clearly. Thoughtful communication is a superpower that sharpens how we collaborate and build • This might not be right if • We're a small team moving fast on hard problems. That might not be a fit if you: • Prefer a slower pace. Customers are depending on what we ship • Enjoy larger organisation structures and staying only within your area of expertise. Taking ownership here means doing whatever the problem needs • Want a traditional engineering team set up, with a predictable roadmap and clearly scoped out tickets provided for you • Aren't comfortable with production responsibility. We're revenue-critical infrastructure - on-call matters here