aws-lambda-power-tuning

Why we included this project

Teams running serverless workloads on AWS often find that Lambda's memory setting quietly affects both cost and latency, and this tool turns that guesswork into a measured experiment. You give it a Lambda function ARN from your account, and a Step Functions state machine invokes that function across a range of power configurations, from 128MB up to 10GB, then reads the execution logs and plots cost against speed for each setting. You can optimize for cost, for speed, or for a balance of the two, so the choice matches your budget or latency targets instead of intuition. It is language agnostic and runs inside your own account, which makes it useful for CPU-heavy jobs, API handlers, and AI inference functions, and parallel execution keeps the whole sweep fast. If you manage a fleet of Lambda functions and want a data-driven answer to how much memory they should use, this is a practical, low-effort way to get it.