Worker Type:
Posting Start Date:
Job Description:
Intern assignment
Describe the responsibilities and activities related to the Assignment. If relevant, relate these activities to the requirements for completion of the Internship as set by the Institute of Education. Include details of department where the Internship will be carried out.
The intern will join the Multi-Modal Generative AI team to work on cutting-edge research and development in collaborative agentic AI frameworks. The assignment will focus on designing, prototyping and evaluating multi-agent systems where specialised AI agents cooperate to reason, plan, use tools, verify outputs and improve solutions across complex Shell workflows.
The intern will explore how frontier AI models and multi-modal capabilities can process diverse inputs such as text requirements, technical documents, diagrams, images, structured data, rules and domain knowledge. The work may involve building agent workflows for task decomposition, decision support, automated verification, optimisation, explainable recommendations and human-in-the-loop components.
The assignment will use state-of-the-art models and frameworks for agent orchestration, tool calling, evaluation and deployment. The intern will collaborate with AI researchers, engineers and domain experts, with expected outputs such as a prototype component, evaluation results, technical documentation and a final demo or presentation.
Goals and objectives
Describe the purpose of the internship, including goals that are preferred to achieve.
The purpose of the internship is to advance Shell’s capability in multi-modal and collaborative agentic AI for complex business, engineering and operational workflows.
This work can bring value to Shell by reducing manual effort, accelerating analysis and decision-making, improving consistency, and making specialist knowledge more reusable across teams and domains. Collaborative agentic frameworks can help Shell move from isolated AI assistants towards scalable AI systems that can plan, verify, optimise and explain their outputs in a controlled and auditable way.
Key objectives include developing and evaluating multi-agent workflows, exploring multi-modal reasoning, improving automated verification and feedback mechanisms, and demonstrating how collaborative agentic AI can support faster, higher-quality and more scalable outcomes across Shell use cases.
-
DISCLAIMER: