Senior AI Engineer

The Role A growing consumer product company is hiring a Senior AI Engineer to move its AI work from proof-of-concept into a genuine production capability. Reporting to the CTO, you'll partner with the business to identify high-value AI use cases, shape the underlying platform and architecture, and take agents from working prototype through to production, then define the standards that keep everything safe as usage scales. This suits someone who has already done hands-on build work and is ready to help set direction for how the whole company applies AI. Why Join * You set the standard, from the first prototype to the platform other engineers build on. * Direct access to senior leadership, shaping AI direction rather than working through someone else's backlog. * Your agents run inside a live consumer product used by real customers. * A fast-moving, scale-up environment with quick decision-making and a strong focus on the end-user experience. About the Company An early-stage but well-established consumer technology company that has grown rapidly and is now profitable, in a competitive digital entertainment/consumer product space. The company has built a product with strong user engagement, centered on social and interactive features that differentiate it from more traditional competitors in its sector, in the way that newer fintech and consumer apps have historically taken share from established incumbents. Requirements *You are:* * AI-first by instinct: your default is to reach for an agent or automated solution before adding headcount or manual process. * Hands-on: you'd rather build the solution than write a strategy document about it. * Direct: willing to say when an idea is flawed, including your own. * Comfortable as the go-to expert without becoming a bottleneck or single point of failure. *Must-have:* * Demonstrable, shipped impact from agentic or LLM-powered systems in production, and the ability to speak candidly about what broke and what you changed as a result. * Hands-on experience with an agent framework (LangGraph, LlamaIndex, Semantic Kernel, ADK, or similar) and production RAG: embedding models, vector stores, re-ranking, and judgment on when live retrieval beats a cached answer. * Strong Python (or equivalent) on solid engineering fundamentals: testing, version control, CI/CD, and building the APIs that serve your own systems. * Hands-on experience with a major cloud provider's managed AI services (Azure/AI Foundry or an AWS/GCP equivalent), plus solid SQL and relational data modeling. * Strong architectural judgment: able to make and defend design trade-offs, and to recognize where an LLM system needs work on cost, latency, or outputs that sound plausible but aren't reliable. * Strong product sense: data-driven, understands genuine user needs, and thinks through second- and third-order effects before shipping. *Nice-to-have:* * Experience proving a system's behavior rather than assuming it: eval sets, output scoring, tracing, regression testing. * Retrieval pipelines at scale, including embedding at volume and maintaining index accuracy as underlying data changes. * Building workflows that hold up under real-world failure conditions: timeouts, failed API calls, human-in-the-loop approval steps. What You'll Do *Be curious* * Work directly with teams to find where an agent would genuinely add value, and get something in front of them quickly enough to learn if the assumption was right. * Choose and justify the build approach (low-code, custom build, or vendor solution) based on cost, control, and speed to value. *Deliver* * Take validated experiments and turn them into production-grade agents that meet the reliability standard a live product demands. * Design explicitly for failure states: timeouts, tool errors, long-running jobs, and points where a decision should be handed back to a human. * Build shared frameworks, templates, and infrastructure so other engineers can ship their own agents without starting from zero. *Own the outcome* * Define the technical shape of the company's agent estate: retrieval approach, evaluation methods, behavioral guardrails, observability, and the path to production. * Decide what requires sign-off before an agent goes live, how it handles user data, and where its authority ends. * Document the rationale behind every agent: what it does, why it was promoted to production, what it costs, and whether it still earns its place. *Collaborate* * Act as an internal AI champion through office hours, demos, and short training sessions. * Partner with department leads to identify where AI genuinely helps, and where it doesn't.