Skip to content

Latest commit

 

History

1,357 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Ravindra K R

DevOps · MLOps · AIOps platform engineer. I run Kubernetes for a fleet of robots, the CI and GPU platform behind it, and the ML training and inference pipelines on top, and I fix what breaks along the way in the open-source tools we depend on.

Golden Kubestronaut (every CNCF and Linux Foundation cloud-native certification)

Open source

Contributions to the projects I operate in production, each fixing something I hit there.

kubernetes/kubernetes ⭐ 128k · mlflow ⭐ 28.1k · argo-cd ⭐ 24.3k · kops ⭐ 16.7k · goreleaser ⭐ 16.1k · cert-manager ⭐ 14.1k · buildkit ⭐ 10.3k · external-dns ⭐ 9.1k · woodpecker ⭐ 7.9k · actions-runner-controller ⭐ 6.5k · tempo ⭐ 5.5k · tetragon ⭐ 5k · opentelemetry-collector-contrib ⭐ 5k · spark-operator ⭐ 3.2k · kueue ⭐ 3k · slam_toolbox ⭐ 2.7k · client_java ⭐ 2.3k · opentelemetry-operator ⭐ 1.8k · opentelemetry-go-contrib ⭐ 1.7k · nativelink ⭐ 1.6k · security-profiles-operator ⭐ 874 · aws-efs-csi-driver ⭐ 806 · percona-xtradb-cluster-operator ⭐ 625 · argocd-agent ⭐ 597 · triggers ⭐ 594 · rclnodejs ⭐ 449 · coroot-node-agent ⭐ 433 · prow ⭐ 327

All contributions
Project Contribution
kubernetes/kubernetes Fix ResourceFieldSelector.Divisor doc: unset or 0 means unscaled
mlflow Reject scheme-less relative workspace default_artifact_root
argo-cd Do not pass --force to server-side apply
kops Cert-manager: add feature gate options for webhook and cainjector
goreleaser Force column.ui=never so tag output is not columnized
cert-manager Certificate-shim: do not mutate the cached object's labels; backported to 1.20, 1.21
buildkit Reject --chown by name with --link at parse time
external-dns Drop AWS SDK dependency from ZoneTypeFilter
external-dns Let a hostname annotation suppress the listener fallback
woodpecker Fix data race in SSE stream handlers when the client disconnects early
actions-runner-controller Add default and per-controller max-concurrent-reconciles flags
actions-runner-controller Document which listenerMetrics example labels are not exposed by default
tempo Cap live trace slice capacity in FindByTraceID
tetragon Pkg/selectors: enforce postfix max length for small BPF programs
opentelemetry-collector-contrib Stop re-emitting records in live mode
spark-operator Stop adding SparkApplication owner reference to executor pods
kueue Automated cherry pick of #15538: Give the leader election lease client its own rest.Config; backported to #15680
kueue Give the leader election lease client its own rest.Config
kueue Use a map for TestSetLeaderElectionConfig cases
slam_toolbox Fix map_start_at_dock & map_start_pose parsing
slam_toolbox Validate save_map name against shell injection
client_java Drop +Inf bound from OpenTelemetry classic histogram boundaries
opentelemetry-operator Add enableServiceLinks to collector and target allocator pod specs
opentelemetry-operator Surface TargetAllocator lookup and build errors on the collector CR
opentelemetry-go-contrib Report Prometheus HTTP server errors in v0.2.0
nativelink Return an error when TlsClient cannot load CA root certificates
security-profiles-operator Harden operator and webhook securityContext: runAsNonRoot, drop ALL
security-profiles-operator Only render SPOd hostPath volumes of enabled features
aws-efs-csi-driver Skip libwrap cleanup when efs state directory does not exist
percona-xtradb-cluster-operator K8SPXC-1919 Use MySQL defaults for replication channel retry count
argocd-agent Process Delete event for app already deleted on the agent
triggers Add a request body size limit to the EventListener sink
triggers Drop cluster-scoped resources from tekton-triggers-aggregate-edit
rclnodejs Add Timer.clock accessor
rclnodejs Write generated/generator.json only after message generation completes
coroot-node-agent Report missing BPF tracing program types clearly
prow Exclude PR from subpool when its changed files cannot be fetched

Generated from the GitHub API: pull requests accepted into repositories with 300+ stars.

What I work on

  • MLOps on EKS: GPU batch and distributed training with Kueue, MultiKueue and Slurm on Kubernetes; Argo Workflows for data and inference pipelines; MLflow for experiment tracking; Karpenter spot capacity for GPUs; KubeRay pilots.
  • CI platform: GitHub Actions runners on EKS with Karpenter, container build caches and registries, benchmark and GPU test suites.
  • AIOps: LLM agents in the operations loop, from incident triage against Prometheus, Loki and Tempo to code changes that only land after mechanical checks.

More on Credly.

About

Building the special repository

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Used by

Contributors

Languages