Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
KubeflowではじめるML on Kubernetes #1 Kubeflowの概要とセッ...
Search
Keita Watanabe
August 17, 2022
500
1
Share
KubeflowではじめるML on Kubernetes #1 Kubeflowの概要とセットアップ
Keita Watanabe
August 17, 2022
More Decks by Keita Watanabe
See All by Keita Watanabe
[re:Invent 2025] Distributed inference on AWS: Deep dive into inference optimizations (AIM353)
keitaw
1
170
Scalable Infrastructure for Large-Scale AI Training with AWS Sagemaker Hyperpod @ Singapore AI Hour
keitaw
0
130
[AWS Summit Japan 2025] Optimizing Foundation Model Development with Amazon SageMaker HyperPod: Insights from Training the Amazon Nova Model
keitaw
0
67
Building foundation models on AWS
keitaw
1
580
[re:Invent2024 Chalktalk] Cost-effectively deploy PyTorch LLMs on AWS Inferentia using Amazon EKS
keitaw
0
150
AWS Summit New York 2024: CMP 301 Demystifying the ML software stack on Amazon EC2 accelerated instances
keitaw
0
510
re:Invent 2023 CMP319 Deploy LLMs with AWS Inferentia & Ray to optimize performance and cost
keitaw
0
73
re:Invent 2023: CMP332 De-mystifying ML software stack on Amazon EC2 accelerated instances
keitaw
1
250
AWS における LLM・GenAI 大規模学習への取り組み / Large scale GenAI・LLM training on AWS
keitaw
1
610
Featured
See All Featured
End of SEO as We Know It (SMX Advanced Version)
ipullrank
3
4.2k
Designing for humans not robots
tammielis
254
26k
Gemini Prompt Engineering: Practical Techniques for Tangible AI Outcomes
mfonobong
2
420
Done Done
chrislema
186
16k
ピンチをチャンスに:未来をつくるプロダクトロードマップ #pmconf2020
aki_iinuma
128
55k
Ruling the World: When Life Gets Gamed
codingconduct
0
240
State of Search Keynote: SEO is Dead Long Live SEO
ryanjones
0
200
Stop Working from a Prison Cell
hatefulcrawdad
274
21k
Money Talks: Using Revenue to Get Sh*t Done
nikkihalliwell
0
240
First, design no harm
axbom
PRO
2
1.2k
AI: The stuff that nobody shows you
jnunemaker
PRO
8
690
職位にかかわらず全員がリーダーシップを発揮するチーム作り / Building a team where everyone can demonstrate leadership regardless of position
madoxten
62
54k
Transcript
Keita Watanabe Kubernetes Meetup Tokyo # 52 LT KubeflowͰ͡ΊΔ ML
on Kubernetes #1 Kube fl owͷ֓ཁͱηοτΞοϓ
Keita Watanabe Machine Learning Solutions Architect ▶ Self-managedͳMachine LearningͷWorkload Λ୲͍ͯ͠ΔSA
▶ લ৬ͰDatascientist/ML Researcherͱͯ͠ ECαΠτ্ͷը૾ݕࡧػೳͷ։ൃʹैࣄ ▶ Twitter: keitaw09 ▶ Linkedin: keitawatanabe ▶ ݱॴଐ: AWS Japan αϯσΟΤΰͰग़ձͬͨτϦ झຯ
Kubeflowͱ Machine Learning Model-Development Lifecycle (MDLC)ͷ ֤εςοϓʢ+ύΠϓϥΠϯʣΛΧόʔ͢ΔToolͷू·Γ KubeflowΛ࠾༻͢ΔϝϦοτ Composability: ༻్ʹԠͯ͡ඞཁͳComponentsΛΓग़ͯ͠༻͍Δ͜ͱ͕Ͱ͖Δ
Portability: Vendor lock-inΛආ͚Δ͜ͱ͕Ͱ͖Δ Scalability: ෛՙʹԠͯ͡Container/NodeΛεέʔϧ͢Δ͜ͱ͕Ͱ͖Δ
Kubeflow overview Central Dashboard Data Ingestion Data Analysis Data Transform
Data Validation Training AutoML Model eval. and valid. Serving Logging KServe Seldon Kube fl ow Pipelines Kube fl ow Notebooks Katib Training Operators MDLCͷ֤εςοϓ ରԠ͢ΔTool
Kubeflow overview Central Dashboard Data Ingestion Data Analysis Data Transform
Data Validation Training AutoML Model eval. and valid. Serving Logging KServe Seldon Kube fl ow Pipelines Kube fl ow Notebooks Katib Training Operators MDLCͷ֤εςοϓ ରԠ͢ΔTool ࠓΔ͜ͱ: Kube fl owͷηοτΞοϓ
Kubeflow setup Amazon EKS্ͰͷKube fl owڥͷߏங • git • kubectl
• eksctl • kustomize • ࠓճAWS Cloud9্Ͱ࡞ۀ͠·͢ cf. https://www.eksworkshop.com/ 020_prerequisites/workspace/ Prerequisites
Demo
Kubeflow setup
To be
࣍ճ…
ࢀߟࢿྉ • https://www.eksworkshop.com/020_prerequisites/workspace/ • https://www.kube fl ow.org/docs/components/ • https://awslabs.github.io/kube fl
ow-manifests/docs/deployment/vanilla/guide/ • https://www.oreilly.com/library/view/kube fl ow-for-machine/9781492050117/