AI Deployment Engineer - Startups

OpenAI OpenAI · AI Frontier · Munich, Germany · Go To Market

AI Deployment Engineer focused on working with strategic startup customers to optimize their AI systems, identify failure modes, and translate learnings into reproducible evaluations and product improvements for OpenAI. This role involves prototyping prompts/agents, designing evaluations, and acting as a technical partner.

What you'd actually do

  1. Work directly with strategic startup customers to understand critical workflows, uncover failure modes, and identify high-impact opportunities for improvement.
  2. Prototype and iterate on prompts, agents, and workflow designs to better understand system behavior and unlock customer value.
  3. Synthesize and deliver valuable feedback to the Product and Research teams, turning real usage patterns into clear, reproducible evals, benchmarks, and technical artifacts that improve model and product quality and ensure customer-grounded learnings influence roadmap and model development.
  4. Build repeatable tools, patterns, and evaluation approaches that raise the quality bar across multiple use cases.
  5. Operate with strong judgment in ambiguous environments, balancing immediate technical problem-solving with longer-term system improvement.

Skills

Required

  • strong software engineering & AI fundamentals
  • experience as a startup CTO, software engineer, ML engineer, Data Scientist or equivalent
  • familiarity with, or interest in, model training pipelines and reinforcement learning
  • experience building AI applications, agents, or evaluation systems
  • ability to reason clearly about model behavior in complex workflows
  • comfortable working directly with highly technical users
  • ability to translate their challenges into concrete technical signals
  • ability to move fluidly between prototyping, debugging, evaluation design, and cross-functional collaboration
  • clear communication across technical and non-technical audiences
  • high agency
  • strong product sense
  • bias toward building durable improvements rather than one-off fixes

Nice to have

  • experience as a technical founder, or engineer at an early stage startup
  • experience shipping production systems end-to-end

What the JD emphasized

  • proficiency in German
  • experience shipping production systems end-to-end is a strong plus
  • experience building AI applications, agents, or evaluation systems

Other signals

  • customer-facing
  • product feedback loop
  • evaluation systems
  • agent/workflow optimization