A practical, end-to-end guide to Hugging Face for the people who run the infrastructure. If you are an admin or platform engineer moving into AI, this series maps every Hugging Face concept onto things you already operate: registries, secrets, storage, scheduling, RBAC, supply-chain security and capacity. Sixteen parts, beginner to production.

1. 01What Hugging Face Actually Is: the Hub, the Libraries, the MapThe Hub is a model/artifact registry — framed against the container registry you already run.
2. 02Navigating the Hub: Models, Datasets, Spaces, and Model CardsA model card is provenance and license metadata to vet before anything enters your environment.
3. 03Accounts, Access Tokens, and the huggingface_hub CLITokens are credentials and secrets: scoping, rotation, safe use in CI.

1. 04Running Your First Model with transformers PipelinesWhat actually loads onto the box and into VRAM when a model runs.
2. 05Tokenizers: What Really Happens to Your TextWhy context limits and throughput are what they are: capacity planning.
3. 06The datasets Library: Loading, Streaming, and StorageData movement, caching and disk, not a data-science detail.
4. 07Model Formats and safetensors: The File Format Is a Security DecisionSupply-chain security: why you block pickle and prefer safetensors.

1. 08Fine-Tuning with Trainer, and LoRA/PEFTA tiny adapter footprint means far less GPU and storage to provision.
2. 09accelerate: Training Across Multiple GPUsMulti-GPU is scheduling, interconnect and capacity — your wheelhouse.
3. 10Pushing Models and Datasets Back to the HubArtifact promotion, versioning and RBAC on a private registry.

1. 11Inference API vs Inference Endpoints vs Self-HostBuild-vs-buy, data egress, and who owns the GPUs.
2. 12Text Generation Inference (TGI) in ProductionA container you deploy, scale and monitor, with a real serving example.
3. 13optimum and Quantization: ONNX, GPTQ/AWQFitting models on the GPUs you actually have, and the cost of not.
4. 14Spaces: Demos with Gradio, and When Not ToWhere a quick demo ends and real platform hosting takes over.

1. 15Security and Governance: Gated Models, Scanning, ProvenanceThe registry and security controls you already run, applied to models.
2. 16Hugging Face in a Private/Enterprise SettingAir-gapped mirroring, the Enterprise Hub, and the bridge to your GPUs.
3. 17Which Model for What: the Model MapText, vision, audio, video: which model family for which task, with sizes and licenses.

New parts are added here as they publish. While the series is being written, this page is your map of what is coming.