What is #Kubeflow?
It's an open-source, Kubernetes-native MLOps platform for developing & deploying distributed ML for the most popular: TensorFlow, PyTorch, and more.
Check out more details on #InfoQ: https://bit.ly/3YXaSBR
#kubeflow #infoq #opensource #kubernetes #MLOps #machinelearning
🚀 Neues Video «Ein Mate mit...» über Machine Learning und #MLOps.
Iwan Imsand, Software Engineer bei Puzzle, nimmt dich in die faszinierende Welt des Machine Learnings mit. Von den Feinheiten zwischen #AI, #MachineLearning und #DeepLearning bis hin zu schicken Open Source Tools, wie: #DVC, #CML, #MLflow, #Kubeflow.
Und weil wir bei Puzzle auch gerne komplexe Probleme lösen, zeigen wir dir, wie wir Unternehmen im MLOps Bereich unterstützen können. 🤖🚀💡
#kubeflow #MLflow #CML #DVC #deeplearning #MachineLearning #AI #MLops
glad to see #MLOps growing within the CNCF. the @kubeflow project has been around since 2017, offering a #cloudnative solution to deploy machine learning workloads on #kubernetes.
Congrats to the #kubeflow team👏
https://cncf.io/blog/2023/07/25/kubeflow-brings-mlops-to-the-cncf-incubator/
#MachineLearning #DevOps
#MLops #cloudnative #kubernetes #kubeflow #machinelearning #devops
Registrations for the fifth edition of the International School on Open Science Cloud are now open #SOSC23 For more details visit: https://web.infn.it/SOSC23 #OpenScience #CloudComputing #DataScience #ArgoWorkflows #kubeflow
#sosc23 #openscience #cloudcomputing #datascience #ArgoWorkflows #kubeflow
HIRING: Senior Machine Learning Ops Engineer / Remote Spain https://ai-jobs.net/J48035/ #AI #MachineLearning #DataJobs #AIcareers #MLjobs #bigdata #DataScience #AIjobs #hiringnow #RemoteJob #Airflow #DeepLearning #DevOps #Engineering #Genetics #Kubeflow #Kubernetes #Mathematics
#mathematics #kubernetes #kubeflow #genetics #engineering #DevOps #deeplearning #airflow #remotejob #hiringnow #AIjobs #DataScience #bigdata #MLjobs #aicareers #DataJobs #MachineLearning #AI
RT igor_os777: Charmed Kubeflow 1.7 is now available http://dlvr.it/Slg6H9 #AIML #AIML #Kubeflow
Good morning folx of #Fediverse :fediverse: ! I interrupt swearing at #CUDA and #Torch to bring you today's #ConnectionList #Introduction #FollowFriday, where I help to more richly connect the Fediverse.
@terrytangyuan is a #MachineLearning #Kubernetes #OpenSource #MLOps engineer who is co-chair of #KubeFlow and maintains #TensorFlow (thank you!).
@yarnspinner is a #dialogue #dialog tool for #GamesDev, brought to you by the team @thesecretlab
@Cerchie is Lucia who is a #DevRel #DeveloperAdvocate at #Confluent, based in the 🇺🇸 👋
@ocramz is Marco and he is into #MachineLearning #ML #NLP #λ #lambda and #NLProc
@lira is into #poetry, #ComputerScience and #Philosophy
@frankjhopwood is a #PhD candidate #researcher #academic at #Groningen working on #speech technology #SpeechSynthesis #TTS for #Frisian. Fun Frisian fact: did you know that "keks" in Frisian means "cookies"?! Cakes and cookies! (Sorry I've always found that very amusing! 🍰 🍪 )
@blairpalese is Director of #Philanthropy at #EthInvest and Media Director and Editor at Climate and Capital Media. She's into #GreenFinance #ClimateRisk #Environment 🌱
Don't forget to share your own #ConnectionList so we can connect more widely ❤️
#fediverse #cuda #torch #connectionlist #introduction #followfriday #machinelearning #kubernetes #opensource #MLOps #kubeflow #tensorflow #dialogue #dialog #gamesdev #devrel #developeradvocate #confluent #ml #nlp #λ #lambda #nlproc #poetry #computerscience #philosophy #phd #researcher #academic #groningen #speech #speechsynthesis #tts #frisian #philanthropy #ethinvest #greenfinance #climaterisk #environment
RT @GOTOcon
Another day, another 2022 GOTO YouTube video on the TOP 5📚
In Kubeflow for ML, @AdiPolak and @holdenkarau talk about the design principles of #Kubeflow & how to use it to build machine learning pipelines.
How did you enjoy this #GOTObookclub episode?
I’m going to be adding tests to our #FluxFramework #Kubernetes operator this week! I’ve seen scorecard for the operator-sdk but I’m looking for examples in the wild using best practices, and possibly not sticking to those suggested by that SDK. The e2e and unit tests in the #kubeflow MPI operator with ginkgo and gomega look nice! 👍 https://github.com/kubeflow/mpi-operator/blob/master/v2/test/e2e/mpi_job_test.go Anyone have other examples they like or experiences to share? 🤔
Dino #koobernottie
#fluxframework #kubernetes #kubeflow #koobernottie
I’m going to be adding tests to our #FluxFramework #Kubernetes operator this week! I’ve seen scorecard for the operator-sdk but I’m looking for examples in the wild using best practices, and possibly not sticking to those suggested by that SDK. The e2e and unit tests in the #kubeflow MPI operator with ginkgo and gomega look nice! 👍 https://github.com/kubeflow/mpi-operator/blob/master/v2/test/e2e/mpi_job_test.go Anyone have other examples they like or experiences to share? 🤔
Dino #koobernottie
#fluxframework #kubernetes #kubeflow #koobernottie
I'm currently keeping the lights on for a deprecated installation of #Kubeflow Pipelines, which uses #Argo Workflows under the hood. None of the components has been updated for at least a couple years now, and we're planning on turning it off in June anyway, but for now there are some critical workflows that depend on it.
Recently the team that manages #Kubernetes updated the #OpenPolicyAgent gatekeeper and added some new policies, which dramatically increased its CPU consumption. The latency of Kubernetes API calls goes through the roof whenever QPS increases, until the Gatekeeper has time to scale out. This causes the Argo workflow controller to fall behind and fail to do whatever it needs to do to terminated pods long enough that the Kubernetes garbage collector gets to them first, which causes the workflow to end in an "error" state.
Now our namespace has been exempted from the Gatekeeper, so all the workflow controller's problems seem to be its own. It was still falling behind and missing terminated pods during periods of high load, so I tried scaling it up by increasing GOMAXPROCS, "workflow-workers", "pod-workers" (a now obsolete concept) as well as "--qps" and "--burst". The last two appear to have been the most important.
Now it's getting into a state where it's repeatedly complaining about conflicts when trying to update workflows. Once it starts happening with a given workflow it keeps happening, but the first retry always works. CPU consumption goes way up when this happens, and it starts missing terminated pods again. Most of the situations I can find like this on the 'net involve something external updating the workflow (we have nothing like that AFAIK) or a mutating admission controller. I can't find any evidence of properties that would be set by a mutating admission controller, so I'm guessing this is a workflow controller bug, probably related to concurrency.
I've reduced "workflow-workers" and "pod-workers" back to 2 and set GOMAXPROCS to 1 (which I think is the default, but I wanted to make sure). It's not happening right now, but load is also low and it seems to happen during periods of high or at least rapidly increasing load.
Thoughts?
#kubeflow #argo #kubernetes #OpenPolicyAgent
Kubeflow is a Cloud Native machine learning toolkit.
#ai #ml #kubeflow #kubernetes #tools
I've been looking at different orchestration tools for #MLOps and was curious about where's the community heading these days.
Of course, GitHub stars are a flawed metric for that, but still... I was curious.
Here's what I found. The main goal was that it needed to be open source. And goes without saying that these tools are not completely alike.
Let me know if there's any tool missing.
#machinelearning #datascience #kubernetes #airflow #flyte #kubeflow #mlflow #argo #polyaxon
#MLops #machinelearning #DataScience #kubernetes #airflow #flyte #kubeflow #mlflow #argo #polyaxon
I live in #Bonn and work in #DataScience and #MachineLearning Engineering. My projects in range from maintaining #Kubeflow installations in on-premises and cloud #Kubernetes clusters to developing #MLOps best practices for teams.
I mostly develop in #Python, #YAML, #Terraform, and am dabbling in #Rustlang.
I got my PhD in #Astronomy being a small part of the #EBHIS and #HI4PI team.
In recent years I’m having fun riding my #eMTB all across the #Eifel and #Ahrtal. #MTB
#introduction #bonn #datascience #machinelearning #kubeflow #kubernetes #MLOps #python #yaml #terraform #rustlang #astronomy #ebhis #hi4pi #emtb #eifel #ahrtal #mtb
The session was presented at the SRE London Meetup https://lnkd.in/eXfCwUDF in #Wipro's Holborn Offices
#Cloud #AI #ML #AIML #MachineLearning #ArtificialIntelligence #Kubeflow #Privacy #DataSovereignty #Security #CyberSecurity #SecOps #MemoryEncryption #UbiquitousDataEncryption #UDE #EndToEndEncryption #Linux #Data #GoogleCloud #GoogleCloudPlatform #GCP #Encryption #DataEncryption #GPU #TPU #Containerization #Anthos #Istio
#wipro #cloud #ai #ml #aiml #machinelearning #artificialintelligence #kubeflow #privacy #datasovereignty #security #cybersecurity #secops #memoryencryption #UbiquitousDataEncryption #ude #endtoendencryption #linux #data #googlecloud #googlecloudplatform #gcp #encryption #dataencryption #gpu #tpu #containerization #anthos #istio
Microsoft Azure users leave front door open for cryptomining crooks - Microsoft has discovered a campaign that exploits Kubernetes to install cryptomining software in i... more: https://nakedsecurity.sophos.com/2020/06/15/microsoft-azure-users-leave-front-door-open-for-cryptomining-crooks/ #cryptocurrency #microsoftazure #cryptomining #kubernetes #tensorflow #microsoft #kubeflow #istio
#istio #kubeflow #microsoft #tensorflow #kubernetes #cryptomining #microsoftazure #cryptocurrency
Kubernetes Falls to Cryptomining via Machine-Learning Framework - Misconfigured dashboards are at the heart of a widespread XMRIG Monero-mining campaign. more: https://threatpost.com/kubernetes-cryptomining-machine-learning-framework/156481/ #maliciouscontainer #misconfiguration #machinelearning #cryptomining #websecurity #containers #kubernetes #microsoft #kubeflow #malware #monero #azure #xmrig
#xmrig #azure #monero #malware #kubeflow #microsoft #kubernetes #containers #websecurity #cryptomining #machinelearning #misconfiguration #maliciouscontainer