Building AI Apps with LLM APIs

About this course

Large language models are now available to any developer through simple APIs. That means one person in Nairobi, Kisumu or Eldoret can build an AI product that would have needed a research team a few years ago. This course shows you how.

You'll make your first API calls, then learn the techniques behind real AI products: system prompts that control behaviour, conversation memory, structured JSON output for reliable automation, tool use so the model can check orders or call your own functions, and retrieval-augmented generation (RAG) so it can answer questions from your own documents with sources.

You'll build and deploy a working chat app with Streamlit, and finish by learning how to keep AI products safe, affordable and trustworthy: evaluation, prompt injection, cost control and data protection under Kenyan law. Examples use the Anthropic Claude API, and the same ideas apply to OpenAI, Google and open-source models.

By the end you will be able to

• Explain how LLM applications are structured, priced and secured
• Call an LLM API from Python and manage API keys safely
• Control model behaviour with system prompts and maintain conversation history
• Get reliable structured JSON output for automation
• Give a model tools so it can look up data and take actions through your code
• Build a retrieval-augmented generation (RAG) system that answers from your documents with sources
• Build and deploy a chat web app with Streamlit
• Evaluate, secure and control the costs of AI applications

Before you start

Python for AI. Machine Learning Essentials and Deep Learning are helpful but not required.