At carVertical, I architected and implemented a production MLOps platform on AWS EKS for GPU-based machine-learning workloads across staging and production. I designed model delivery, GPU-aware scheduling, dynamic provisioning, networking, observability, deployment safety, and automated recovery.
I build end-to-end software delivery workflows using trunk-based development, GitHub Actions, Docker, Amazon ECR, Helmfile, and reusable Helm charts. I also built inference infrastructure with NVIDIA Triton Inference Server, NVIDIA GPU Operator, KAI Scheduler, Karpenter, Argo Rollouts, Kyverno, and Envoy Gateway.
Previously, at Cloudvisor, I re-engineered a Ruby on Rails data-processing workload in Python and Pandas, reducing execution time from more than 24 hours to approximately 3.5 hours. At Yapily, I operated Kubernetes across AWS and Google Cloud and introduced GitOps-based deployments with ArgoCD.
Across more than 20 years in software, infrastructure, security, and automation, I’ve also led DevOps teams, taught Python and DevOps, and mentored engineers through practical production scenarios.
