Applied AI Engineering

Overview

Large language models have changed what software can do — but shipping AI features that are useful, safe and reliable is an engineering discipline. This program teaches that discipline.

You will build retrieval-augmented applications, agents that use tools, and the evaluation harnesses that tell you whether a change made things better or worse. Throughout, you'll practise responsible AI: privacy, safety and honest communication of limitations.

What you'll be able to do

• Integrate large language models into applications through their APIs
• Build retrieval-augmented generation over your own documents
• Design agents that call tools safely and predictably
• Measure quality with automated and human evaluation
• Identify and mitigate risks such as data leakage and prompt injection

1. 01How language models workA practical mental model of tokens, context, sampling and cost.2 weeks

   • Tokens and context windows
   • Sampling and determinism
   • Cost and latency trade-offs
2. 02Building with LLM APIsStructured outputs, streaming and robust error handling.3 weeks

   • Prompt design patterns
   • Structured outputs
   • Streaming responses
   • Handling failures

   Project: An AI study assistant with structured answers

3. 03Retrieval-augmented generationGround answers in your own data with chunking, embeddings and ranking.4 weeks

   • Embeddings
   • Chunking strategies
   • Hybrid search and re-ranking
   • Citations and grounding

   Project: A question-answering system over course documents

4. 04Agents and tool useLet models take actions — with guardrails.3 weeks

   • Tool definitions
   • Planning loops
   • Permissions and human approval

   Project: A scheduling agent with safe tool calls

5. 05Evaluation and safetyKnow whether your system works — and when it fails.2 weeks

   • Building eval sets
   • Automated graders
   • Red-teaming and prompt injection
6. 06CapstoneShip an evaluated AI product.2 weeks

   • Proposal and eval plan
   • Build and iterate
   • Demo day

Tools you'll use

• Python
• Jupyter
• Vector databases
• FastAPI
• Evaluation frameworks

• AR

  Ayesha Rahman

  Instructor, Applied AI

  Machine-learning engineer focused on reliable, evaluated AI systems.

  • Python
  • LLM applications
  • Evaluation
  • Data engineering

Questions about this program

More answers in our general FAQ.