Omkar Satish Khadake
About Candidate
DevOps Engineer with 2+ years of experience running production systems on AWS, Azure, Kubernetes, and on-premises NVIDIA GPU infrastructure. HashiCorp Certified Terraform Associate.
Key strengths: Infrastructure as Code (Terraform, AWS CDK, CloudFormation), GitOps and CI/CD (Argo CD, GitHub Actions with OIDC), Kubernetes (EKS, KEDA autoscaling), and observability (Grafana, Prometheus, Loki, Tempo, OpenTelemetry).
Highlights: Built an AI gateway on AWS Bedrock that led to a paid client engagement and cut AI provider costs by 10-20%. Replaced long-lived cloud credentials with Azure OIDC federation. Wrote reusable Terraform modules for nine Azure services and delivered a DEV-to-PROD migration.
Looking for DevOps, Platform, and Cloud Engineer roles. Notice period: 2 months.
Location
Education
GPA: 8.22/10
Work & Experience
Spectra (AI video analytics, AWS EKS + NVIDIA DGX): operated an EKS cluster with seven microservices; implemented Argo CD GitOps deployments, KEDA autoscaling and off-hours shutdown to cut idle compute; integrated two GPU inference services on an on-premises NVIDIA DGX server; built a configuration-driven provider factory for AWS, Azure and GCP; deployed Prometheus, Loki and Tempo with five Grafana dashboards, Teams/email alerts and k6 release tests.
Enterprise AI Gateway (AWS Bedrock, MLflow, FastAPI, OPA, Redis): built a proof of concept that led to a paid client engagement; moved authentication to Entra ID SSO, removing API keys from developer workstations; cut AI provider costs by 10-20% with Redis quotas, spend limits and usage reporting; implemented a default-deny policy layer with JWT validation and OPA model allowlists.
SmartHub (Azure, Terraform, Container Apps, GitHub Actions): created reusable Terraform modules for nine Azure services; replaced long-lived CI credentials with Azure OIDC federation and approval-gated Terraform apply; delivered the DEV-to-PROD migration across five workstreams.
Deployed and operated Frappe and ERPNext on Linux with Nginx, Redis and Supervisor-managed workers; scripted backups, restores and upgrades across staging and production. Delivered 5+ custom modules, 20+ DocTypes and a document management system that reduced retrieval time by 40%; ERPNext process changes reduced manual fee errors by 40%.