Senior Software Engineer at Red Hat | Kubernetes Contributor
Platform engineer working on Kubernetes infrastructure, focusing on Dynamic Resource Allocation for GPU/accelerator scheduling, AI inference platform enablement, and node-level platform configuration. Contributing to upstream Kubernetes, CNCF projects, and OpenShift platform components.
Contributing to upstream Kubernetes DRA implementation and downstream OpenShift enablement for GPU multi-tenancy.
Upstream (kubernetes/kubernetes):
- KEP-5004 (DRAExtendedResources): upgrade/downgrade e2e tests (#136568), race condition fix in quota test (#139085)
- Global cache for DeviceClass-to-extended-resource mapping (#134326)
- DRA integration test fixes (#137432)
- KEP-5004 test planning for rollout/upgrade/rollback (kubernetes/enhancements#5751)
Downstream (OpenShift):
- Enabled DRA featuregate by default in OpenShift API (openshift/api#2498)
- DRA e2e tests for NVIDIA GPU hardware (openshift/origin#30758)
- KEP-4815 Partitionable Devices e2e tests (openshift/origin#31230)
- CI infrastructure for DRA validation on NVIDIA GPU
Core contributor to GPU slicing operator enabling multiple AI/ML workloads to share GPU hardware through fine-grained MIG-based resource allocation. Shipped as Tech Preview in OpenShift 4.18-4.21.
- Operator maintenance: CVE remediation, libnvidia-ml updates, hardening
- Konflux release engineering for OCP 4.20-4.21: release configuration, NVIDIA CUDA/RHEL-AI repository enablement, RPM validation fixes, FBC catalog pipeline management
- E2E testing infrastructure across KIND, OpenShift SNO, and multi-GPU clusters
- Repository: openshift/instaslice-operator
Building a Claude Code plugin for automated DRA validation on OpenShift clusters.
- Repository: openshift-eng/ai-helpers#520
Drove node-level platform features across Machine Config Operator, Cluster Node Tuning Operator, CRI-O, and OpenShift API.
- Cgroups v1 to v2 migration: Built controller support for cgroup mode switching and led the platform-wide migration (2022-2026)
- Cgroups v1 deprecation: Removed cgroupv1 code paths across MCO, NTO, and OpenShift API (MCO#5399, NTO#1428, API#2579)
- Worker Latency Profiles: Node configuration controller for latency-sensitive edge computing deployments
- Evented PLEG: CRI-O and kubelet integration for event-driven Pod Lifecycle Event Generator, replacing the polling-based approach
170+ merged PRs across upstream Kubernetes, CRI-O, and OpenShift platform components (GitHub only).
Member of: kubernetes | kubernetes-sigs | openshift | cri-o
| Repository | Focus |
|---|---|
| kubernetes/kubernetes | DRA e2e testing, KEP-5004, performance improvements |
| kubernetes/enhancements | KEP-5004 test planning |
| kubernetes-sigs/jobset | DRA integration documentation |
| openshift/origin | DRA & GPU e2e testing |
| openshift/instaslice-operator | GPU slicing operator |
| cri-o/cri-o | Evented PLEG, CI improvements |
| Event | Topic |
|---|---|
| KubeCon + CloudNativeCon India 2026 (Mumbai) | Multi-Tenancy of AI Inference Workloads on OpenShift using llm-d and DRA — Red Hat Booth Demo |
| KubeCon India 2025 | InstaSlice (DAS) - GPU Slicing for AI/ML Workloads — Red Hat Booth Demo |
Go Kubernetes OpenShift DRA NVIDIA MIG llm-d vLLM GPU Operator CRI-O Helm Kustomize Prow Tekton Konflux GCP
- GitHub: @sairameshv
- LinkedIn: linkedin.com/in/sai-ramesh-vanka
- Email: [email protected]
Building production infrastructure for Kubernetes platforms and contributing to the cloud-native ecosystem.
Last updated: July 20, 2026