FSD-2 Agentic Development and Intelligent Applications

Building on the AI integration competencies developed in the first year, this course focuses on designing and implementing agentic applications that combine language models, external tools, retrieval mechanisms and controlled workflows to perform multi-step tasks. Students learn models and techniques such as retrieval-augmented generation, embeddings, tool calling, planning and reasoning, multi-agent patterns and context engineering, and design AI features with attention to user experience and operational costs. They plan and build an agentic application with human oversight, in line with principles for responsible AI such as fairness, transparency and explainability.

In the second half of the course, students test what they have built. They assess its quality and reliability, detect hallucinations and unsafe behaviour, defend against prompt injection, and add guardrails and automated evaluation. They also assess their AI features against requirements for security, data protection and AI regulation.