How would you implement auto-scaling groups, load balancing, and serverless architectures for compute and scaling at Epsilon?
