
Build a cost-tracking script for AI infrastructure
You'll have a working script that pulls your actual cloud spending on compute and flags when costs spike beyond projections.
What you'll make
AI Cost Monitor Script
- fetch_bills() — pulls current spending from cloud APIs: what to include
- detect_spikes() — flags unusual cost increases: what to include
- project_usage() — models next 90 days based on trends: what to include
- alert_thresholds() — sends notifications when limits are hit: what to include
- main() — orchestrates daily monitoring: what to include
Paste your working Python script with example output from your actual cloud bills
The 5 parts
Each part ends with a check you can see.
Separate Real from Circular Costs
Analyze your current AI spending to identify usage-driven costs vs relationship-subsidized pricing
Done when
A categorized breakdown of your last 90 days of AI infrastructure spending with circular vs direct costs identified
Build the Data Pipeline
Write fetch_bills() to pull current spending from your cloud provider's API
Done when
A working fetch_bills() function that pulls your last 30 days of AI-related charges with line-item detail
Build Spike Detection
Write detect_spikes() to flag unusual cost increases that might indicate runaway workloads
Done when
A working detect_spikes() function that catches cost increases above your defined threshold with test cases from your historical data
Model Usage Trends
Write project_usage() to forecast costs based on your actual usage patterns
Done when
A working project_usage() function that forecasts next 90 days of costs with confidence intervals based on your historical data
Deploy the Complete Monitor
Integrate all functions into a working monitoring script and run it on your infrastructure
Done when
A complete working script that runs daily monitoring, sends alerts for spikes, and provides cost projections with at least one week of real monitoring data
Capstone: you put together your AI Cost Monitor Script.
Where this helps
An elective: a single lesson that sits outside rtwo's job certificates and doesn't count toward one.
About this lesson
What you'll be able to do
- Run a working cost-tracking script on your actual cloud bills
- Set automated alerts for cost spikes above your defined thresholds
- Distinguish between real AI infrastructure costs vs circular revenue flows in your organization
- Have a 90-day cost projection model based on your actual usage patterns
Who it's for
- Engineering managers tracking AI infrastructure spend
- Finance teams monitoring cloud costs for AI workloads
- Technical leads building cost-aware AI systems
- Anyone running AI workloads who's seen surprise bills
What it won't cover
- Building production billing systems from scratch
- Tax optimization or accounting policy
- Vendor negotiation strategies
- Multi-cloud cost allocation across teams
Skills you'll practise
- Cloud billing API integration
- Cost anomaly detection
- Usage pattern analysis
- Threshold-based alerting