Advanced AI Prompt for GPU-Optimized EKS Cluster Design for AI/ML

Act as an AI/ML Infrastructure Engineer. Architect a specialized EKS cluster optimized strictly for high-velocity Deep Learning and LLM training workloads utilizing heavy [GPU_INSTANCE_TYPE] nodes (e.g., `p4d.24xlarge`). Provide the Terraform node group configuration mandating the use of the Amazon EKS accelerated AMI containing the exact [NVIDIA_DRIVER_VERSION]. Apply rigid taints (`nvidia.com/gpu=true:NoSchedule`) to prevent standard microservices from burning expensive GPU compute. Detail the DaemonSet installation of the NVIDIA Device Plugin [K8S_DEVICE_PLUGIN_VERSION] to expose the hardware to the Kubelet. Output a sample TensorFlow `Job` manifest targeting the [NAMESPACE] that explicitly requests `limits: nvidia.com/gpu: 4` and mounts high-speed Amazon FSx for Lustre storage.