Skip to content

Before You Begin

Choose Your Path

This content is available in two formats:

Path Description Best For
Workshop Hands-on, self-paced learning with detailed step-by-step instructions Individual learners working through the material on their own
Demo Presentation-ready format with ACME Corporation business scenario Instructors, presenters, or those following along with a live demo

Both paths cover the same 5 modules but with different contexts:

  • Workshop: Focus on learning vLLM Playground features at your own pace
  • Demo: Business-focused narrative showing how ACME Corporation transforms their customer support with AI

Timing and Schedule

📚 Workshop Path (90 minutes)

Self-paced, hands-on learning with detailed step-by-step instructions.

Module Topic Duration
Module 1 Getting Started with vLLM Playground 18 minutes
Module 2 Advanced Inferencing: Structured Outputs 18 minutes
Module 3 Advanced Inferencing: Tool Calling 18 minutes
Module 4 Advanced Inferencing: MCP Integration 18 minutes
Module 5 Performance Testing 18 minutes

🎬 Demo Path (45 minutes)

Presentation-ready format for live demonstrations.

Module Topic Duration
Module 1 Getting Started with vLLM Playground 8 minutes
Module 2 Advanced Inferencing: Structured Outputs 10 minutes
Module 3 Advanced Inferencing: Tool Calling 10 minutes
Module 4 Advanced Inferencing: MCP Integration 10 minutes
Module 5 Performance Testing 7 minutes

Quick Start (20-36 minutes)

If you're short on time, focus on these modules:

Path Modules Duration
Workshop Module 1 + Module 5 36 minutes
Demo Module 1 + Module 5 15 minutes

Technical Requirements

Software Versions

Component Version
vLLM Playground v0.1.1+
vLLM v0.11.0+
Podman 4.0+ (or Docker)
Python 3.10+ (required for MCP)
NVIDIA GPU CUDA support
Web browser Chrome, Firefox, Safari, Edge

Environment Access

After installing vLLM Playground, you will have access to:

Resource URL
vLLM Playground Web UI http://localhost:7860
vLLM API Endpoint http://localhost:8000

Environment Setup

Quick Setup

# 1. Install vLLM Playground
pip install vllm-playground

# 2. Pull the GPU container image (~10GB)
vllm-playground pull

# 3. Start the playground
vllm-playground

# 4. Open http://localhost:7860 in your browser

What's Included

Component Status
vLLM Playground CLI ✅ Installed via pip
Podman/Docker ⚠️ Required (install separately)
NVIDIA GPU drivers ⚠️ Required for GPU mode
Python 3.10+ ⚠️ Required for MCP support

Optional Packages

The following packages can be installed for additional functionality:

Package Install Command Purpose
MCP Client pip install mcp or pip install vllm-playground[mcp] Required for Module 4
GuideLLM pip install guidellm or pip install vllm-playground[benchmark] Required for Module 5

Pre-install Everything

To install all optional dependencies at once: 

pip install vllm-playground[mcp,benchmark]

Setup Validation

Run these commands to verify your environment:

# Verify vLLM Playground installation
vllm-playground --help

# Verify Podman (or use 'docker version')
podman version

# Verify GPU availability (if using GPU)
nvidia-smi

# Check Python version (for MCP)
python3 --version

# Verify MCP installation (optional - for Module 4)
python3 -c "import mcp; print('MCP installed successfully')"

# Verify GuideLLM installation (optional - for Module 5)
guidellm --help

Troubleshooting Guide

Common Setup Issues

Problem Solution
vllm-playground: command not found Verify the installation path is in your PATH, or reinstall with pip install vllm-playground
Permission denied when running Podman Ensure rootless Podman is configured, or use sudo
NVIDIA driver not found or GPU not detected Install NVIDIA drivers and verify with nvidia-smi
Container image pull fails Check network connectivity and container registry access
Port 7860 already in use Run vllm-playground stop or use --port to specify a different port

During Workshop Support

# Check vLLM Playground logs
vllm-playground status

# View container logs
podman logs vllm-service

# Restart if needed
vllm-playground stop
vllm-playground

Follow-up Resources

After the Workshop

Additional Learning Paths

Level Focus Area
Intermediate Explore different model architectures and their tool calling capabilities
Advanced Deploy vLLM Playground on OpenShift/Kubernetes for enterprise scale
Production Implement custom MCP servers for your specific use cases

Authors and Contributors

Primary Author: Michael Yang
Last Updated: January 2026
Workshop Version: 1.0

Feedback and Questions:

Ready to Start?

Choose your learning path: