Senior Agentic Platform Engineer

You’ll join VaultSpeed’s small Agentic team as a hands-on senior engineer, building AI agents that customers rely on in production. The platform runs on Omnigent, an open-source agent platform that VaultSpeed extends for enterprise use, with a graph-based context store and Data Vault expertise packaged into the agents as skills.

The work spans the whole stack: the agents and their skills, the platform services they run on, the Kubernetes infrastructure behind them and the interface people use. You’ll own areas of the system from technical design to release, and your decisions will shape how a new product grows.

• Build and improve the agents that ingest sources and generate data models and code, and the skills that carry data modelling knowledge, deciding what belongs in a prompt, a skill, a tool or plain deterministic code
• Build evaluations and regression suites, so changes to prompts, models and tools are judged on evidence
• Extend the platform layer on top of Omnigent: multi-tenancy and tenant isolation, authentication, the MCP tool suite, the context store, the agent marketplace and the LLM gateway
• Keep the Omnigent fork in sync with upstream, resolve conflicts with customisations and contribute fixes back to the open-source project
• Work on the Kubernetes sandbox infrastructure where agents run (templates, warm pools, persistence, suspend and resume, observability, cost) and ship changes through GitOps with Argo CD
• Investigate agent failures through logs and model traces, and fix root causes rather than symptoms
• Write technical proposals, review designs and pull requests, and support releases

• 5+ years of professional software engineering, including owning the technical design of significant components through to production
• Hands-on experience shipping LLM-based features to real users (beyond prototypes): tool calling, structured outputs, context management
• Strong back-end fundamentals: services and APIs, authentication, data modelling, databases
• Docker and hands-on experience deploying and debugging services on Kubernetes through a GitOps workflow
• Comfortable working in a large codebase you don’t own and resolving non-trivial merge conflicts
• Strong analytical thinking: you get from a vague symptom to a root cause, also in non-deterministic systems
• Daily use of AI coding tools such as Claude Code or Cursor
• Experience working in a Scrum team across a distributed setup
• Fluent English

Nice to have

• MCP, the Claude Agent SDK, LangGraph or similar agent frameworks
• LLM evaluation and observability tools such as MLflow or Langfuse
• Graph databases (Neo4j)
• Multi-tenant SaaS and identity: OIDC, SSO, WorkOS or similar
• Kubernetes agent-sandbox or other sandboxed execution environments
• Data engineering or data warehousing
• Maintaining a fork of, or contributing to, an active open-source project
• Comfortable making minor front-end changes

• Building an AI product in a well-defined domain, where agent output can be checked against a real model
• Joining a new product early, in a small team where your design decisions shape the platform
• Remote work in a team organised around trust and outcomes
• Private health insurance
• Company events