Technical practice
How would you integrate AI Foundation services into a product such as Delfi or Lumi?
What to coverThe posting names FM Hub, Model Ops, AI Workspace, Agent Workspace and GenAI Infrastructure as services to integrate. Suggested preparation: explain how you would define service boundaries, connect a product workflow and verify that the integration is ready for production.
What makes a GenAI integration pattern reusable across product teams?
What to coverThe role calls for reusable APIs, components and integration patterns. Suggested preparation: discuss stable interfaces, configuration and the trade-off between a shared component and a product-specific implementation.
How would you deploy a domain-specific foundation model in a product pipeline?
What to coverThe posting requires model integration and deployment with domain-aligned behavior and performance. Suggested preparation: describe how you would connect the model to domain inputs and outputs, validate its behavior and plan for model changes.
How would you define a benchmark with geoscience domain experts?
What to coverThe role includes jointly defining benchmark datasets, evaluation metrics and acceptance criteria. Suggested preparation: explain how representative cases and domain judgments would inform the benchmark, including how you would handle gaps or disagreements in the data.
How would you make model evaluation continuous through a release process?
What to coverThe posting calls for continuous evaluation of quality, robustness and domain relevance, embedded in development and release work. Suggested preparation: describe when evaluations would run, how results would be compared and what evidence would support an acceptance decision.
How would you evaluate inference efficiency without losing domain performance?
What to coverThe role includes optimizing deployment workflows for inference efficiency and production readiness. Suggested preparation: discuss how you would measure performance and cost alongside domain quality, then explain the trade-offs behind a deployment choice.
How would you design a multi-agent reasoning and tool-use workflow?
What to coverThe posting calls for production-grade agentic solutions involving multi-agent orchestration, reasoning pipelines and tool use. Suggested preparation: explain why you would divide responsibilities among agents or tools, how information would pass between them and where failures would be detected.
How would you combine foundation models, domain data and simulation tools?
What to coverThe role describes compound AI systems for geoscience and energy challenges. Suggested preparation: choose a clearly identified illustrative workflow and explain how each element would contribute, including how you would check outputs against domain constraints.
How would you evaluate an end-to-end domain agent?
What to coverThe posting explicitly calls for agent evaluation frameworks, separate from evaluating a foundation model alone. Suggested preparation: cover task completion, tool-use errors and robustness across representative domain cases, and explain how failures would be traced to a component.
What would you need to learn before applying AI to seismic interpretation or reservoir simulation?
What to coverThese applications are listed as preferred experience, while familiarity with geoscience, subsurface or energy workflows is required. Suggested preparation: explain how you would work with domain experts to map inputs, decisions and validation criteria rather than assume that a general model metric establishes domain usefulness.
Behavioural practice
Tell me about a time you translated a domain problem into a technical solution.
What to coverThe role expects translation of geoscience or energy problems into agentic solutions. Suggested preparation: use a real situation, explain how you established the domain need and chose an approach, then state the result without overstating your contribution.
Tell me about a time you coordinated an integration across teams.
What to coverThe posting names product teams, domain experts and the AI Foundation Team as collaborators. Suggested preparation: describe the situation, your actions to resolve dependencies or differing priorities, and the resulting integration outcome.
Tell me about a time you agreed on evaluation criteria with subject-matter experts.
What to coverThe role requires collaboration on benchmarks, metrics and acceptance criteria. Suggested preparation: give a real example of the initial disagreement or uncertainty, how you reached a workable definition and what resulted.
Tell me about a time you helped others adopt a technical capability.
What to coverThe posting calls for demos, reference implementations, workshops, training and documentation. Suggested preparation: describe the audience and situation, what you created or taught, and the observable result.
Tell me about a time you improved a system for production use.
What to coverThe role emphasizes deploying production-grade agentic systems and improving inference efficiency and readiness. Suggested preparation: describe a real production constraint, the trade-off you made, your actions and the measured or observed result.