How One Engineer Reduced AWS Costs by 70 Percent with Spot Instances

In this episode, Lucas and Luna explore how a single engineer at a mid-sized fintech company slashed their AWS bill by 70 percent—over $2 million annually—by migrating stateless workloads to spot instances. They break down the technical strategy: designing for interruption tolerance using checkpointing and distributed queues, setting up a spot-fleet mix across instance types and availability zones, and automating fallback to on-demand when spot capacity was unavailable. Lucas explains the key architectural changes needed, from decoupling state from compute to implementing graceful shutdown handlers. Luna shares why many teams avoid spot instances due to fear of interruptions, and how proper design turns that risk into a major cost advantage. The episode also covers the one gotcha that nearly broke the migration: a hidden AWS service quota on spot instance requests per region. Listeners will walk away with a concrete framework for evaluating whether spot instances fit their workloads and a step-by-step approach to making the switch safely. This is episode 29 of The Software Engineering Podcast with Fexingo. #AWS #SpotInstances #CloudCostOptimization #Fintech #Engineering #DevOps #Infrastructure #CloudComputing #CostSaving #Resilience #Checkpointing #DistributedSystems #Technology #SoftwareEngineering #FexingoBusiness #BusinessPodcast #TechPodcast #CloudArchitecture Keep every episode free: buymeacoffee.com/fexingo