About this ebook
AI models don't scale themselves. Kubernetes does.Training a model on your laptop is the easy part. Getting that model to serve a million predictions a day — reliably, efficiently, and without bankrupting your infrastructure budget — requires a different skill set entirely. Kubernetes has become the universal answer to that challenge, and every serious AI team in production is running on it.This book teaches you to deploy, scale, and operate AI workloads on Kubernetes at a professional level — from your first GPU Pod to multi-cloud LLM serving infrastructure.Inside you will find:The complete Kubernetes foundation for AI — architecture, core concepts, GPU node pool setup on GKE, EKS, and AKS, NVIDIA device plugin configuration, and the containerization patterns that make ML workloads reproducible and portable across every environmentML model deployment — the Deployment, Service, and Ingress pattern for production inference, GPU scheduling with node affinity and taints, GPU sharing with MIG and time-slicing, and purpose-built serving frameworks including NVIDIA Triton Inference Server and Seldon CoreBatch and training workloads — batch inference with Kubernetes Jobs and indexed parallelism, single-node and distributed training configurations, checkpoint patterns for spot instance resilience, and priority classes for training workload managementMLOps with Kubeflow — notebooks, PyTorchJob and MPIJob training operators, PyTorch DDP distributed training, Horovod multi-node training, Kubeflow Pipelines workflow orchestration, and Argo Workflows DAG pipelinesScaling, storage, and observability — HPA with custom inference metrics, VPA for right-sizing, Karpenter for GPU node auto-provisioning, persistent storage for model artifacts, NVIDIA DCGM GPU metrics, Prometheus and Grafana dashboards, and KEDA scale-to-zero for cost controlProduction infrastructure — Pod Security Standards for AI workloads, Secrets management, NetworkPolicy for model isolation, multi-cloud federation with Liqo, Istio multi-cluster service mesh, and the complete production AI platform architecture with ML CI/CD pipelines and model registry integrationLLM deployment — vLLM high-throughput serving, tensor parallelism for 70B+ parameter models, PagedAttention memory optimization, FP8/INT4 quantization, and OpenAI-compatible API gateway patternsEvery chapter includes working YAML manifests and code you can apply to your own cluster immediatelyThe AI teams deploying at scale are all running on Kubernetes. This book is the fastest path to joining them.By Fatima Ahmad. Available on Google Play Books.
AI models don't scale themselves. Kubernetes does.Training a model on your laptop is the easy part. Getting that model to serve a million predictions a day — reliably, efficiently, and without bankrupting your infrastructure budget — requires a different skill set entirely. Kubernetes has become ...
Discover more
Rate this ebook
Tell us what you think.
Reading information
Smartphones and tablets
Install the Google Play Books app for Android and iPad/iPhone. It syncs automatically with your account and allows you to read online or offline wherever you are.
Laptops and computers
You can listen to audiobooks purchased on Google Play using your computer's web browser.
eReaders and other devices
To read on e-ink devices like Kobo eReaders, you'll need to download a file and transfer it to your device. Follow the detailed Help Center instructions to transfer the files to supported eReaders.
