SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Proxima Fusion, Europe's fastest-growing and best-funded fusion company, is seeking a Senior Agent Platform Engineer to own the technical architecture and operation of its AI agent platform. This role is critical to scaling autonomous agents from early-stage product into reliable, governed infrastructure used across the company by physicists and engineers.
You will be the technical owner responsible for designing, building, and operating the agent platform runtime and its core primitives: durable sessions, sandboxed execution environments, long-running task execution, memory management, and tool integration. Your work will enable agents to operate across knowledge bases, write and execute code, run simulations, and support complex engineering workflows.
Key responsibilities include: architecting the platform to turn a rapidly evolving agent ecosystem into stable building blocks; co-designing integration layers connecting agents to internal and external systems; implementing safety and governance features including permission flows, capability boundaries, audit logging, and protections against prompt injection; building evaluation and observability layers for non-deterministic systems; and operating the platform on Google Cloud with production discipline.
You will work directly with demanding technical users—scientists and engineers—gathering feedback and shaping the platform roadmap in collaboration with the Enterprise AI Team Lead. You'll need to evaluate emerging agent harnesses, models, and protocols, deciding what becomes a supported primitive versus a passing experiment.
Required: 5+ years software engineering experience with deep expertise in Python and distributed backend systems. You must have architected and operated production-grade agentic or LLM systems beyond prototype phase, with hands-on knowledge of the current agent ecosystem (MCP, agent SDKs/CLIs, multi-agent orchestration, sandboxed execution). Production experience running stateful, long-running services on Kubernetes and a major cloud platform (ideally GCP) is essential. You treat evaluation, observability, and security as design constraints, think in platforms and internal products, and communicate with clarity and low ego.
Nice-to-have skills include experience with model gateways, scientific/HPC computing, open-source AI/agent tooling contributions, and technical mentorship track record.
Proxima Fusion combines stellarator optimization, advanced computation, machine learning, and high-temperature superconducting magnets to develop the world's first commercial stellarator fusion power plant. The company is backed by over €650M and operates across Munich, Zurich, and Oxford.
