Gopala Krishna Akkineni · DevOps engineer · Bloomington, IL
I turn risky releases into routine ones.
I build the pipelines, clusters and GPU platforms that get software and machine learning models to real users. At State Farm that has meant releases 40% faster, 95% fewer interruptions and 99.9% uptime, with every environment written down as code.
Read my résumé Watch a deploy happen
Day to day
- Kubernetes
- Terraform
- GitHub Actions
- ArgoCD
- AWS
- GCP
- Docker
- Prometheus
- Python
Open to DevOps, SRE and platform engineering roles. Based in Bloomington, IL, with State Farm since March 2025.
- 40%
shorter release cycles after rebuilding the delivery pipeline
- 95%
fewer interruptions during releases with blue-green and canary deploys
- 30%
faster model training and inference on autoscaling GPU clusters
- 99.9%
uptime on infrastructure I provisioned with Terraform and CloudFormation
What I do
Most of my week falls into three kinds of work. Each one links to a story with the details.
Build the delivery path
CI/CD on GitHub Actions, Jenkins and Buildkite, with every environment defined in Terraform or CloudFormation.
Faster releasesRun the platform
EKS and GKE clusters, GitOps with ArgoCD and FluxCD, and GPU capacity for model training and LLM apps.
GPU platform for MLKeep it safe and visible
Prometheus, Grafana and ELK for seeing problems early, and Vault, Trivy and Snyk so bad builds stop in CI.
Seeing problems firstSelected work
Each one is written up properly: the situation, what I built, the decisions behind it and what changed.
- 40%Faster path to productionRebuilt CI/CD across GitHub Actions, Jenkins and Buildkite
- 95%Releases nobody noticesBlue-green and canary rollouts with GitOps on EKS and GKE
- 30%GPU platform for MLAutoscaling GPU and TPU clusters for training and LLM apps
- 25%Seeing problems firstPrometheus, Grafana, CloudWatch and ELK with alerts that matter
- 3Security in the pipelineSecret managers behind one approach, and scans that fail builds
- 5A monitoring stack with tested alertsTested alert rules, one command to start the stack
- 3A canary release that stops itselfCanary steps, tested on a real cluster with a broken release
- 0A GPU cluster that costs nothing when idleGPU nodes running until a job needs one
- $0This websiteA static site run like a production service
On GitHub
Pulled live from my GitHub, so this list keeps itself up to date.
More projects are on the way. This site is the first one, and you can read its source.
Right now
Updated November 2026.
- Working on the platform behind State Farm’s AI and ML work.
- Building DevOps projects in public on GitHub. This site is the first one.
- Looking for my next DevOps, SRE or platform role.
Hiring for DevOps or platform work?
I’m open to DevOps, SRE and platform engineering roles. Email is the fastest way to reach me.