Everything you need to plan, deploy, operate, secure and optimize VMware Private AI Foundation with NVIDIA on VCF 9, in one complete, sequential series. Thirty parts, from first principles to production day-2 and the field topics that come after. Start at Part 1, follow a reading path below, or jump to the part you need.
Series complete · 30 of 30 parts published
Reading paths: where to start by role
Architect
Design the platform end to end: concepts, reference architecture and the trade-offs that bite.
Operator (Day 2)
Run it in production: deploy, monitor, troubleshoot and upgrade without breaking the GPUs.
Decision map: jump to the call you need to make
Which GPU should I buy?Part 5: L40S vs H100 vs H200 vs Blackwell
vGPU, MIG or passthrough?Part 6: GPU partitioning
How do I serve and scale models?Part 25: NIM Operator reference architecture
Fine-tune or use RAG?Part 27: NeMo Customizer
Is Private AI even the right platform?Part 30: PAIF vs OpenShift AI vs hyperscaler
Core Path: the end-to-end spine
Phase 1 · Foundations
- 01What VMware Private AI Foundation with NVIDIA actually is
- 02Architecture & components, end to end
- 03Licensing & editions: PAIF add-on and NVIDIA AI Enterprise
Phase 2 · Plan & Design
- 04Planning & prerequisites before you touch a GPU
- 05GPU hardware selection for Private AI
- 06vGPU, MIG and passthrough: how to slice a GPU
- 07Reference architecture & sizing a first cluster
Phase 3 · Deploy
- 08Preparing the VCF workload domain for AI workloads
- 09Installing the NVIDIA GPU Operator & vGPU drivers
- 10Deep Learning VMs: provisioning & model development
- 11Deploying NIM microservices for inference
Phase 4 · AI Services & Apps
- 12Private AI Services: Model Store & Model Runtime
- 13Vector database with Data Services Manager (pgvector)
- 14Building a RAG pipeline on Private AI Foundation
- 15Agent Builder & agentic applications
- 16Self-service provisioning with VCF Automation
Deep Dives: specialized tracks
Phase 5 · Operate (Day 2)
Phase 6 · Secure & Disconnected
Phase 7 · Optimize & Master
- 21Performance tuning & benchmarking inference
- 22Model lifecycle & MLOps on private cloud
- 23Troubleshooting VMware Private AI Foundation
- 24Upgrades, 9.0 to 9.1, and what’s next
Extended Field Topics: from the engagements
Phase 8 · Field Topics
- 25NVIDIA NIM Operator: declarative model serving by reference architecture
- 26Networking for AI workloads: segmentation, ingress and east-west
- 27Fine-tuning with NeMo Customizer: LoRA, full SFT and when to bother
- 28Guardrails & responsible AI: what NeMo Guardrails actually stops
- 29Disaster recovery & multi-tenancy: what to protect and how to share
- 30PAIF vs OpenShift AI vs hyperscaler managed AI: an honest verdict
Want the concepts first? For a vendor-neutral, ground-up explanation of how generative AI works, before building it on this stack, see my Generative AI: From Zero to Mastery series.
