One-day developer course · Sydney, Melbourne and online
Building cost-efficient LLM applications
A one-day, hands-on course on the economics of AI features. You start with a working application and a bill, measure what each request costs, build an evaluation set, then apply each savings lever in turn and verify that quality holds. You leave with the same application costing a fraction of what it did at nine o'clock.
Who it's for
Engineers who own the AI line on the cloud bill.
Software engineers
Who built an AI feature that works and now need it to be affordable at scale.
Technical leads
Who need to set cost budgets for AI features and review designs against them.
Platform and FinOps engineers
Who manage cloud spend and need visibility into token-based costs.
Technical founders
Who need AI features priced into a product with a margin.
What you'll be able to do
By the end of the day you will be able to:
- Attribute token costs to features, customers and request types from logs and traces
- Build an evaluation set and use it to verify every cost change
- Route requests to the smallest model that meets the quality bar
- Use prompt caching, batch processing and context trimming correctly for your provider
- Tune retrieval so it returns less and answers better
- Set token, step and spend limits on agents, with alerts in your own accounts
Agenda
A full day, mostly labs on one application that gets cheaper hour by hour.
- 9:00
Where the money goes
Token pricing across providers, input versus output, and reading a provider invoice. Lab: instrument the application and attribute cost per feature.
- 10:15
Evaluation first
Building an evaluation set from real examples and automated grading. Lab: baseline the application's quality before changing anything.
- 11:30
Model routing
Choosing models per task, cascades and fallbacks. Lab: route three request types and measure cost and quality.
- 13:30
Caching, batching and trimming
Prompt caching, batch APIs, system-prompt and context reduction. Lab: apply each and re-run the evaluation.
- 14:45
Retrieval and agents
Chunk counts, re-ranking, and limits on agent steps, tokens and tool calls. Lab: fix an expensive RAG pipeline and a looping agent.
- 16:00
Dashboards and forecasting
Cost-per-outcome dashboards, budget alerts and a 12-month forecast. Review of each participant's own application.
What's included
You keep everything you build.
- The lab application, before and after, with its evaluation suite
- Cost attribution and dashboard templates for Langfuse, Helicone and CloudWatch
- A cost review checklist for LLM features
- Cloud and model credits for the labs
- Slides, reading list and a certificate of completion
- A 30-minute cost review of your own application within a month
Questions
Which providers does the course cover?
OpenAI, Anthropic, Amazon Bedrock, Azure OpenAI and Google Vertex AI. Caching and batching differ by provider, and we cover each.
Do we need our own usage data?
No. The lab application comes with realistic traffic. In in-house sessions we can work on your real logs instead.
How is this different from the AI cost audit?
The AI cost audit is us measuring and planning savings for your system. The course teaches your engineers to do it themselves, and to build the next feature efficiently from the start.
Is there a non-technical version?
The cost and budgeting material is covered for leaders in AI for business leaders.
Related courses
Securing LLM applications
For developers. Prompt injection, data exfiltration, agent permissions and the OWASP LLM Top 10, with hands-on attack and defence labs.
Course detailsHalf 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 detailsOne 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 detailsMake the next AI feature affordable from day one
Public cohorts run monthly in Sydney, Melbourne and online. Register your interest and we will send the next dates.