Tool or Technology
TT0175

Hugging Face in Action

Discover, Fine-Tune & Deploy AI Models

Hugging Face
Duration
16 Hours
Level
Intermediate
Delivery
Instructor-Led
Format
Capability Training

Programme Objectives

  • Develop functional proficiency in discovering, evaluating, using, and managing AI models and datasets through the Hugging Face ecosystem.
  • Use Transformers and Hugging Face libraries to integrate pretrained language, vision, audio, and multimodal models into applications.
  • Prepare datasets and adapt pretrained models using practical fine-tuning and parameter-efficient techniques.
  • Deploy AI applications using Spaces, Inference Providers, and dedicated Inference Endpoints.
  • Build reusable AI and agent workflows using Hugging Face Agents, MCP, Skills, model repositories, and production-oriented governance practices.

Technology covered

Hugging Face Hubhuggingface_hubTransformersDatasetsTokenizersInference ProvidersInferenceClientInference EndpointsSpacesGradioAutoTrain AdvancedPEFT / LoRAHugging Face AgentsHugging Face MCP ServerAgent Skills

Who should attend

  • AI & Machine Learning Engineers
  • Data Scientists
  • Generative AI Developers
  • Software Developers
  • Python Developers
  • NLP Engineers
  • Computer Vision Engineers
  • MLOps Engineers
  • AI Application Developers
  • Data & AI Consultants
  • Technical Leads
  • Solution Architects & AI Platform Teams

Prerequisites & Participant Readiness

  • Basic proficiency in Python programming
  • Basic understanding of machine learning and Generative AI concepts
  • Familiarity with datasets, training, inference, and model terminology is beneficial
  • Basic knowledge of APIs and JSON is recommended
  • Familiarity with Jupyter Notebook, Google Colab, or another Python environment is helpful
  • Basic Git and repository knowledge is beneficial
  • Access to a Hugging Face account and suitable development environment should be available for hands-on exercises
  • GPU access may be useful for selected model-training activities but is not mandatory for the complete programme
  • Completion of shorter Hugging Face programmes is not required

TOC Modules

Concepts
  • Understanding Hugging Face as an AI model, dataset, application, inference, and collaboration ecosystem
  • Understanding the relationship between Hub, Transformers, Datasets, Spaces, and inference services
  • Understanding pretrained models versus task-specific and fine-tuned models
  • Understanding local, serverless, dedicated, and hosted-model workflows
  • Understanding the Discover → Evaluate → Use → Adapt → Deploy lifecycle
Practical activities
  • Creating and configuring a Hugging Face account and access token
  • Navigating Models, Datasets, Spaces, and Collections
  • Identifying appropriate Hugging Face resources for a defined AI business requirement
  • Comparing local and hosted approaches for the same model task

Scenarios

Enterprise Knowledge Assistant

Business Documents → Dataset Preparation → Embedding Model Selection → Semantic Retrieval → Language Model → RAG Workflow → Gradio Interface → Hugging Face Space / Endpoint → User Validation

Participants build a source-grounded knowledge assistant using models from the Hugging Face Hub, semantic retrieval, and a deployable interface while validating unsupported questions and model limitations.

Domain Model Adaptation & Deployment

Business Requirement → Model Discovery → Model Card & License Review → Training Dataset → AutoTrain / PEFT-LoRA → Evaluation → Hub Repository → Inference Endpoint → Application Integration

Participants select an appropriate pretrained model, adapt it to domain-specific data using an efficient fine-tuning approach, evaluate its performance, and prepare it for controlled application deployment.

Related programmes

Continue with programmes from the same capability area.

Take the next step

Ready to make this programme work for your team?

Customise modules, duration and business scenarios for your team.

Instructor-ledVirtualHybrid

Designed around your roles, tools and real workflows.