AssemblyAI runs speech-to-text inference on EC2 Spot Instances
Start here
AssemblyAI runs its GPU-heavy speech-to-text inference on Amazon ECS using a mix of EC2 Spot and On-Demand Instances, spanning the g4dn, g5 and g6 GPU families in US West (Oregon) and Europe (London). To keep service commitments when Spot capacity is reclaimed, an Amazon SQS-based retry mechanism returns interrupted jobs to the queue through an EventBridge-triggered Lambda handler. According to the AWS Machine Learning Blog, this setup preserves SLA commitments while lowering costs.
Generated from reports · updated 36 min ago
Report timeline
Follow the reports to see the event from different sides.
- AWS Machine Learning BlogAssemblyAI uses Amazon EC2 Spot Instances to cost-optimize speech-to-text inference on AWS
AssemblyAI runs its GPU-heavy speech-to-text inference on Amazon ECS with a mix of EC2 Spot and On-Demand Instances, spanning g4dn, g5, and g6 GPU families across US West (Oregon) and Europe (London). An Amazon SQS-based retry mechanism returns interrupted jobs to the queue via an EventBridge-triggered Lambda handler, preserving SLA commitments while cutting costs.
Heat for this event
Not enough continuous observations yet to draw a trend.