Cloud Native AI Day Europe 2024
2024-03-19
Paris, France
as the organisers put it
Cloud Native AI Day Europe has officially wrapped! Thank you to all the attendees who joined us. Watch keynotes and all breakout sessions on Cloud Native AI Day’s YouTube channel.
| title | author | length |
|---|---|---|
| Welcome + Opening Remarks | Yuan Tang, Rajas Kakokdar | 09m |
Opening remarks lay out concrete logistics and focal projects: 15 talks 9:15–17:00 with a panel at noon and a reception at 17:30, live captioning and… | ||
| Training and Optimisation of Large Transformer Models: An ATLAS and CERN Use Case - Ricardo Rocha, CERN & Maxence Draguet, University of Oxford | Ricardo Rocha, Maxence Draguet | 37m |
CERN runs a Gen2 Transformer for ATLAS jet flavour tagging on a Kubeflow CERN ML cloud, using Kubeflow + Katib for distributed training and hyperparameter… | ||
| Gen-AI at Scale: Simplifying Orchestration of Healthcare Applications Across Multi-Cluster Environme | Selvi Kadirvel, Christopher J Nuland | 27m |
Pipeline uses AWS SQS triggers and KEDA autoscaling to start embedding jobs with Hugging Face and LangChain, stores vectors in Weaviate vector DB (also tested… | ||
| Lightning Talk: Best Practices for LLM Serving with DRA | Chen Wang, Abhishek Malvankar | 10m |
Serving LLMs on A100/H100 shows cluster GPU utilization as low as 20–30% and per-token generation around 50 ms per-token under test loads. Experiments show… | ||
| Unleashing Kubernetes Intelligence - running k8sgpt utilizing your own fine-tuned LLM | Mario Fahlandt | 12m |
Claims a working on‑prem pipeline that fine‑tunes an LLM and plugs it into k8sgpt so cluster diagnostics run locally (no cloud LLM calls). Uses Kubeflow… | ||
| Pods Everywhere! InterLink: A Virtual Kubelet Abstraction Streamlining HPC Resource Exploitation | Diego Ciangottini | 32m |
An open source extension of Virtual-Kubelet called interLink exposes a Kubernetes API that maps Pod requests to remote runtimes via an Interlink API server and… | ||
| Lightning Talk: Cloud Native Networking for AI : Strengthen CNI for RDMA | Weizhou Lan, Junnan Shi | 09m |
An open source CNI called SPID extends macvlan and SR-IOV to provide unified RDMA-friendly networking with IPAM, clusterIP access, multi-interface routing and… | ||
| Panel: Beyond the Clouds: Charting the Course for AI in the CloudNative World | Alolita Sharma, adhuri, Ricardo Aravena, Rajas Kakodkar, Cathy | 35m |
CNCF working group releases a white paper (≈20 pages) and commits to a targeted reference architecture and a CloudNative AI landscape that uses OCI registries… | ||
| The Hitchhiker's Guide to Kubernetes Platforms: Don’t Panic, Just Launch! | Alexa Nicole Griffith, Tessa Pham | 26m |
Three concrete platform APIs power the system: debugging API that returns a single status summary with customized dashboard links and translated pod errors… | ||
| Efficient Multi-Cluster GPU Workload Management with Karmada and Volcano | Kevin Wang | 27m |
Karmada and Volcano combine to run gang-scheduled GPU jobs across clusters using Prometheus metrics, a scheduler estimator that simulates in-cluster… | ||
| Make Descheduler Smarter and Safer: How We Apply Reinforcement Learning in Descheduling Strategies | Xuming Wang, Haosong Huang | 28m |
Don't miss out! Join us at our next Flagship Conference: KubeCon + CloudNativeCon North America in Salt Lake City from November 12 - 15, 2024. Connect with our… | ||
| Effortless Scalability: Orchestrating Large Language Model Inference with Kubernetes | Rohit Ghumare, Joinal Ahmed | 23m |
Orchestrates deployment of large open models like Falcon and Llama 2 on Kubernetes using Kubernetes CRDs and a built in HTTP inference server, packaging model… | ||
| Resource-Aware Scheduling for Production GenAI with RAG running on Multicluster Cloud Kubernetes | Anne Marie Holler, Dave Southwell | 26m |
Claims a reproducible reference architecture that runs GenAI+RAG across multiple Kubernetes clusters using a policy-driven cluster scheduler to place and… | ||
| Building Serverless AI Apps with Spin and WebAssembly | Matt Butcher, Radu Matei | 25m |
Spin compiles small serverless handlers to WebAssembly so a 2.4 MB wasm module running a LLaMa2 Chat LLM starts in ~0.51 ms locally, swaps transparently from… | ||
| Scale Your Batch / Big Data / AI Workloads Beyond the K8s... | Antonin Stefanutti, Anish Asthana | 29m |
Runs a 1,000-job experiment (each job with 10 pods, 3–5 minute work, 15 minute deadline) on Quark-simulated nodes and shows the default Kubernetes scheduler… | ||
| Closing Remarks | Rajas Kakokdar | 01m |
Highlights practical use cases: LLMs for networking, scheduling, and integrating AI in Kubernetes with concrete community focus on graduating, incubating and… | ||
| nothing matches | ||