Senior DevOps Engineer at Invisible Technologies in London

Job description

About Invisible Invisible is an Applied AI research company closing the gap between frontier models and the enterprise. We provide the platform infrastructure that makes AI work in production at scale: unify fragmented data, map complex processes, orchestrate and evaluate agents, and stand up hyper-personalized workflow applications fast for companies like Grubhub, Alliance Global, and Thomson Reuters. Where models fall short, we build RL environments and data to fix them. The result: a virtuous cycle between enterprise deployment and model training. Recognized as Best Predictive Analytics Platform of the Year at the 2026 AI Breakthrough Awards, Invisible makes AI work. About The Role As a Senior DevOps Engineer, you'll own a domain of the Invisible platform — such as identity, compute, delivery, or agent infrastructure — and set the direction for how it works. You'll define the patterns that engineering teams across the company adopt, and be accountable for their reliability, security, and cost. This is a role where judgement matters more than throughput. Deciding which problems are worth solving, recognising when the unremarkable solution is the correct one, and knowing when to decline a bespoke request from a client engagement are the day-to-day of the job. We are looking for engineers with a first-principles understanding of platform practice, not just fluency in it. Code review, staged rollouts, change management, and least privilege were all designed around human engineers working at human speed. Those assumptions no longer hold cleanly. A meaningful part of this role is determining what still applies, what needs reworking, and what has to be built from scratch. What You’ll Do Own a platform domain end to end, including architecture, roadmap, reliability, security posture, and cost Lead design and trade-off discussions across performance, scalability, cost, security, and reliability, and document the decisions Build multi-tenant infrastructure that client engagements can self-serve onto without the Platform team becoming a delivery bottleneck Advance our agent-native engineering model — safe execution environments, credential and secret handling, internal tooling and MCP servers, and evaluation of where agents genuinely improve outcomes Set standards for how the engineering organisation works with coding agents, including review practice, context engineering, and what should and shouldn't be automated Partner with client-facing engineering teams to understand their infrastructure requirements and translate them into reusable platform capability Mentor engineers on the team, review their designs, and raise the technical bar without absorbing their work Own on-call and post-incident review practice What We Need Kubernetes & Cloud: Proven experience designing and operating production Kubernetes at multi-team scale, on GCP, Azure, or AWS Infrastructure as Code: Deep experience with Terraform, Helm, and GitOps delivery, including a clear view of how these degrade when implemented poorly Platform Security: Strong grounding in identity and SSO, workload identity, secrets management, and network policy. You can reason about a threat model directly rather than through a compliance checklist Programming: Strong Python. You build tooling, not only configuration AI Coding Agents: Exceptional practical command of agentic coding tools, and a track record of making other engineers more effective with them. You've built the harnesses, guardrails, and internal tooling, and you have a considered view of where agents deliver value and where they quietly cost more than they save Engineering Judgement: A first-principles grasp of why platform practices exist. You can identify which survive contact with agentic engineering, which need reworking, and which were only ever ceremony Communication: Excellent written communication. You can take an ambiguous problem and produce a document that makes the decision clear to technical and non-technical