Dmitriy K.

I cut AWS bills and build infrastructure that a small team can actually run. On an AI SaaS I brought the monthly cloud bill down ~30% and cut EKS node startup from 15 to 2 minutes. Senior DevOps engineer with 10+ years, AWS first, with hands-on Azure and GCP. Since 2020 I work mostly on AI products: GPU workloads on EKS, LLM serving, AWS Bedrock. I've built infrastructure from scratch as the only DevOps engineer on the team, so I'm used to owning the whole setup and writing it down for the next person. Some results: - ~30% lower monthly cloud cost with Karpenter for on-demand GPU nodes, Spot Instances, rightsizing and KEDA autoscaling - EKS node startup from 15 to 2 minutes with a custom Packer AMI with pre-baked container images - AWS platform for a financial analytics product, all in Terraform: EKS, ECS, RDS, Redshift, WAF, and the LLM layer on Bedrock with least-privilege IAM - Kubernetes on AWS, Azure and GCP (EKS, AKS, GKE) managed with Terraform, Helm and GitHub Actions - Supported SOC 2 compliance audits What I can help with: - AWS cost review and optimization: compute, Graviton, Spot, Savings Plans, NAT and data transfer, logs and storage - New AWS setup or taking over an existing one, everything in Terraform - Kubernetes (EKS) setup, upgrades and autoscaling with Karpenter and KEDA - CI/CD with GitHub Actions, GitLab CI or Bitbucket Pipelines, with safe releases and rollback - Monitoring with Prometheus, Grafana, Datadog and CloudWatch - Infrastructure for AI/LLM products: GPU nodes, vLLM, Bedrock, SageMaker Stack: AWS, Terraform, Terragrunt, Kubernetes, Helm, Karpenter, KEDA, Docker, GitHub Actions, GitLab CI, Jenkins, Prometheus, Grafana, Datadog, HashiCorp Vault, PostgreSQL, Redis, Python, Bash. Short tasks like a cost review and long-term work are both fine. Send me what you have and I'll tell you where I would start.