AI Engineer

## About the role At Gradient Labs, we're on a mission to make exceptional customer service the norm. Founded in 2023, we've quickly gone from an idea to a growing team with customers you know (and probably love). Our AI agent helps businesses handle even the trickiest, high-stakes customer support queries safely and effectively, all while giving them the visibility and control they need to trust the outcomes. We're a small but mighty team of builders from leading companies like Monzo, Pleo, and Google. We work in a hybrid model from our London office, a short walk from Liverpool Street Station, where we collaborate and connect 2-3 days a week. If you're excited to tackle some of the hardest problems in AI and help shape the future of customer operations, we'd love to hear from you. ## Responsibilities This is a build-and-ship role. You'll turn ambiguous customer support problems into reliable, observable AI agents that handle live conversations for real users. You'll work close to production — designing prompts and tool flows, building eval suites, shipping changes, watching what breaks, and iterating fast. - Build and operate AI agents in production: Design, implement, and maintain agentic systems powered by LLMs — handling tool calling, multi-step reasoning, and integration with customer APIs and data sources. You'll own these systems end-to-end: reliable, observable, and auditable from day one. - Translate business problems into agentic workflows: Work directly with enterprise customers to understand their workflows, surface the highest-leverage automation opportunities, and frame them as well-scoped agent problems with clear success criteria. You'll be the technical counterpart in customer conversations, turning ambiguity into a concrete plan. - Build robust evaluation infrastructure: Create and maintain eval suites drawn from real-world scenarios and edge cases. Go beyond vibes-based testing: structured evals measuring accuracy, safety, and latency, tied to clear business outcomes, used to drive systematic improvements to prompts, tools, and behaviour. - Enhance our agent: Develop, evaluate, and optimise the skills that make up our agent. Curate datasets, iterate on improvements, test changes, and ship successful approaches into production. - Shape our internal AI platform: Contribute to shared libraries, patterns, and standards for how we build, evaluate, and deploy agents across customers. Help define how we approach prompting, tool orchestration, retrieval, and monitoring. - Experiment and prototype: Keep up with the latest in NLP, agentic systems, and generative AI. Prototype against our hardest problems with a bias toward shipping experiments quickly rather than long research cycles. - Analyse data: Work across customer queries, support tickets, and related data to find patterns and identify what our agents could automate next. - Drive technical decisions: Scope your own work, push back when the framing is wrong, and tell us when the plan needs to change.