Describe a severe production outage or bottleneck you resolved on AWS.

How to Explain in an Interview:

• Problem: During a high-traffic flash sale, our serverless API Gateway + Lambda service spiked to 500 errors and p99 response times surged to 8+ seconds.
• Root Cause:
  1. Lambda concurrency hit our regional account quota (1,000 concurrent executions), causing incoming invocations to be throttled.
  2. Lambdas inside a VPC lacked VPC Endpoints and routed all S3 and DynamoDB calls through a saturated single NAT Gateway, causing connection timeouts.
• Fix:
  1. Requested a quota increase and configured Provisioned Concurrency on checkout Lambdas to eliminate cold starts.
  2. Added S3 and DynamoDB Gateway VPC Endpoints directly to the subnet route tables, bypassing the NAT Gateway and routing AWS traffic through AWS's free private internal backbone.

# Terraform: Gateway VPC Endpoints for S3 and DynamoDB (Zero NAT fee, sub-ms latency):
resource "aws_vpc_endpoint" "s3_gateway" {
  vpc_id          = aws_vpc.main.id
  service_name    = "com.amazonaws.us-east-1.s3"
  route_table_ids = aws_route_table.private[*].id
  vpc_endpoint_type = "Gateway"
}