Kubeflow Summit Europe 2024
as the organisers put it
Kubeflow Summit Europe has officially wrapped! Thank you to all the attendees who joined us. Watch keynotes and all breakout sessions on Kubeflow Summit Day’s YouTube channel.
who it's for
MLOps engineers, machine learning engineers, platform engineers, data scientists, site reliability engineers, security engineers, engineering managers
topics
Kubeflow Pipelines v2; KServe model serving; Training Operator LLM SDK; Katib hyperparameter tuning; distributed PyTorch DDP; DeepSpeed model sharding; GitHub Actions CI for notebooks; model registry alpha; rootless containers proof; MinIO artifact injection; OpenShift AI integration; Spark Operator donation
| title | author | length |
|---|---|---|
| Welcome + Opening Remarks | Ricardo Rocha, Amber Graner | 09m |
Kubeflow announces it joined CNCF as an incubation project last July and is actively working through the graduation checklist, including an explicit push on… | ||
| A Scalable Platform for Training and Inference Using Kubeflow | Philipp Gadow, Diana Gaponcic | 29m |
CERN runs a Kubeflow-based platform that manages the full ML lifecycle from GitLab-built container training images to notebooks, TensorBoard and Triton… | ||
| Transforming Data Science at PepsiCo: The Kubeflow Revolution | Milos Grubjesic, Chris Addy | 31m |
About 15 months of work replaces isolated notebooks and ad hoc scripts with a Kubeflow-based ML platform: a single monorepo for shared code, a component… | ||
| Driving AI Innovation the Open Way | Jeremy Eder | 09m |
Red Hat announces it rejoined Kubeflow and commits engineers to concrete pieces rather than vague goals: a model registry (over 40% of survey respondents… | ||
| Lightning Talk: Improve Data Science Collaboration with CI/CD! | Urvashi Mohnani | 05m |
Automated CI/CD uses GitHub Actions to detect notebook adds/updates, builds container images on notebook change using Buildah/Podman/Docker under the hood… | ||
| Lightning Talk: Kubeflow Security Working Group Update 2024 | Julius von Kohout, Diana Dimitrova Atanasova | 11m |
Security working group reports concrete changes for Kubeflow 1.9: swap the OIDC/off-service with an of2 proxy to enable token‑based machine‑to‑machine auth… | ||
| Efficient Integration of Kubeflow Tools: An ISS Data Case Study on Anomaly Detection | Henrik Steude, Christian Geier | 23m |
Kubeflow Pipelines v2 orchestrates Dask for parallel preprocessing, Katib for hyperparameter search, PyTorch Operator for multi‑node training and KServe for… | ||
| The Good, the Bad, and the Missing Parts of Kubeflow | Ricardo Martinelli de Oliveira | 24m |
User survey shows top usage is pipelines, notebooks, and KServe and many users deploy from raw manifests, causing upgrade and install pain; pipelines v2 has… | ||
| AutoML & Training Working Group Updates | Andrey Velichkevich, Yuki Iwai, Johnu George, Amber Graner | 36m |
Roadmap items: Katib targets V1 API graduation, advanced parameter distributions like log-uniform, and native LLM hyperparameter tuning; Training Operator adds… | ||
| Lightning Talk: Leveraging Kubeflow for Next-Generation AI Healthcare Solutions | Alin Spinu | 14m |
Roche describes a Kubeflow-based mlops platform that actually runs drug discovery predictors at scale using Kubeflow, Knative, Helm and Argo CD, plus an… | ||
| ⚡ Lightning Talk: Unleash the Power of Generative AI Using Only Open Source Technolgies | Christian Kadner | 08m |
Deploys open source LLMs on small Kubernetes clusters using KServe + TGIS + Caikit and runs them in production on Red Hat OpenShift. Shows that model sharding… | ||
| Closing Remarks | Amber Graner | 02m |
Spark Operator is now part of Kubeflow and a project registry is becoming a reality with slides and links promised for follow up; Kubeflow completes its first… | ||
| nothing matches | ||