Two-day developer course · Sydney, Melbourne and online
Building with large language models: a hands-on course for developers
Two days on what it takes to put an LLM application into production: prompts that hold up, retrieval over your own documents, tool use and agents, evaluation, security and cost control. You leave with a working assistant deployed on AWS.
Who it's for
Working developers who need to ship LLM features, not just try them.
Software engineers
Who've been asked to add AI to a product and want to do it properly the first time.
Data scientists and analysts
Who want to move from notebooks to deployed, monitored applications.
Technical leads
Who need to review LLM designs and set standards for their team.
Technical founders
Who are building an AI product and want to avoid the expensive mistakes.
What you'll be able to do
Skills you can apply in your own codebase the following week.
- Write prompts and structured outputs that behave predictably in production
- Build retrieval-augmented generation (RAG) over your own documents, with citations
- Give models tools and build simple, reliable agents
- Create an evaluation set and measure accuracy on every change
- Defend against prompt injection and data leakage
- Control cost and latency, and deploy on AWS with monitoring
Agenda
Mostly hands-on. Each topic is a short explanation followed by a lab that builds towards a complete assistant.
- Day one: foundations and retrieval
- 9:00
How LLMs behave
Tokens, context windows, temperature and why the same prompt gives different answers. Choosing a model.
- 10:00
Prompts that hold up
System prompts, examples, structured outputs and validation. Lab: an extraction service that returns valid JSON every time.
- 11:30
Retrieval over your documents
Chunking, embeddings, hybrid search and re-ranking. Lab: a question-answering service with citations.
- 14:00
Evaluation
Building a test set, automated grading and regression tests for prompts. Lab: score your assistant and improve it.
- 16:00
Debugging retrieval
Why RAG systems give wrong answers and how to find out which part failed.
- Day two: agents and production
- 9:00
Tool use and agents
Function calling, planning loops and when an agent is the wrong answer. Lab: an assistant that looks things up and takes actions.
- 11:00
Security and guardrails
Prompt injection, data leakage, personal information and access control. Lab: attack your own assistant, then defend it.
- 13:30
Cost and latency
Caching, model routing, batching and streaming. Measuring cost per request.
- 14:30
Deploying on AWS
Amazon Bedrock, Lambda, logging and monitoring. Lab: deploy your assistant with a budget alert.
- 16:00
Review and next steps
Design review of each participant's own use case.
What's included
You keep everything you build.
- A complete, deployed assistant and its source code
- A starter repository with evaluation, logging and deployment set up
- Cloud and model credits for the labs (you don't need your own AWS account)
- A production readiness checklist for LLM features
- Slides, reading list and a certificate of completion
- A 30-minute design review of your own project within a month
Common questions
Which programming language is used?
Labs are available in Python and TypeScript. Choose whichever you're more comfortable with.
Which models and providers does it cover?
The concepts apply to any provider. Labs use Amazon Bedrock with a choice of models, and we show the differences when using the Anthropic and OpenAI APIs directly.
Do I need an AWS account?
No. We provide a sandbox account and credits for the course. You'll also learn how to set up the same thing in your own account.
Can we use our own documents?
In public sessions we use realistic sample data. In in-house sessions we can work on your own documents inside your cloud account.
Related courses
AWS fundamentals for small teams
The 20% of AWS that covers 90% of what a growing company needs, including the billing alerts that stop nasty surprises.
Course details →Half day · $290 per seatAI for business leaders
What today's AI can and can't do, how to find real opportunities, and how to manage the risk. No code.
Course details →One day or more · From $3,900 per dayCustom in-house training programs
Any course, or a mix, delivered to your team on your systems and your data. Up to 12 people per session.
Course details →Ship your first LLM feature properly
Public cohorts run monthly in Sydney, Melbourne and online. Register your interest and we'll send the next dates.