Principal Platform Engineer, Agentic AI Infrastructure

Design and build the platform infrastructure for AI agents to run scientific modeling workflows under controlled policies. You will make computations reproducible and verifiable, build the services agents use to access scientific tools, and own the engineering needed to release and operate the system across different deployment environments.

• Make the scientific compute core reproducible through pinned R, Python, and solver tool chains, content-addressed container images, software bills of materials, and re-execution checks.
• Build the agent harness, including tool routing, policy controls, per-tenant sandboxing without network egress, usage metering, and signed provenance records for results.
• Provide a versioned API and agent-facing tools with a shared contract, plus packages that let users re-run results outside the platform.
• Operate CI evaluations for conformance, re-execution, and regression, retaining results by procedure version.
• Own release engineering and operations: CI/CD, signed releases, installation qualification, environment management, monitoring, incident response, and deployment procedures.
• Implement technical controls supporting information security and AI management standards.

• At least five years of platform, infrastructure, or backend engineering, including senior ownership of a production system.
• Practical experience with reproducible builds and execution, dependency pinning, artifact hashing, immutable versions, and software supply chain integrity.
• Strong Python skills and comfort with R and scientific computing tool chains.
• Experience with containers, infrastructure as code, production CI/CD, Kubernetes or a similar orchestrator, observability, and release management.
• Experience designing versioned APIs used by external consumers, and with workflow or DAG runtimes and AI agent tool-calling systems.
• Security engineering experience with multi-tenant isolation, secrets, and authorization; familiarity with GxP or equivalent regulated software, or willingness to learn it.

• Experience with hermetic build systems.

Flexible hours, a home office policy, and regular team gatherings across Europe. The role offers ownership of a technically significant platform problem and the opportunity to shape infrastructure for certified scientific software.