Vertex AI Multimodal LLMs Training in Copenhagen, Denmark

The Vertex AI Multimodal LLMs Training is a hands-on program focused on building applications that use multimodal large language models on Google Cloud Vertex AI, enabling systems that understand and generate across text, images, audio, and video.

The course covers architecture, API usage, prompt design, workflow orchestration, and production deployment for enterprise-grade multimodal AI solutions.

Introduction to Multimodal AI on Vertex AI

• What is multimodal AI?
• Enterprise use cases for multimodal systems
• Vertex AI multimodal model ecosystem
• Benefits and limitations

Vertex AI Multimodal Architecture

• Multimodal model endpoints
• Data flow for multimodal apps
• Authentication and access control
• Cost and scalability considerations

Text + Image Understanding

• Image captioning and description
• Visual question answering
• OCR and document understanding
• Image classification and tagging

Image Generation & Editing

• Text-to-image generation
• Image variations and inpainting
• Style and brand control
• Quality and resolution handling

Audio & Speech with Multimodal LLMs

• Speech-to-text integration
• Text-to-speech generation
• Voice-based interactions
• Accessibility use cases

Video Understanding & Generation

• Video summarization and tagging
• Scene and object understanding
• Storyboard-style generation
• Video pipeline workflows

Multimodal Prompt Engineering

• Designing cross-modal prompts
• Aligning text + visual context
• Controlling outputs across modalities
• Iterative refinement strategies

Multimodal Workflow Orchestration

• Chaining multimodal model calls
• Combining vision + language + audio
• Conditional and branching workflows
• State and context management

Structured Outputs & Validation

• Enforcing structured multimodal outputs
• JSON schemas for multimodal tasks
• Validation and error handling
• Reliability patterns

Safety, Moderation & Policy Controls

• Content safety for images, audio, and video
• Moderation APIs and filters
• Brand safety and compliance
• Responsible AI for multimodal content

Performance, Cost & Scaling

• Latency optimization
• Batch vs real-time processing
• Caching and reuse
• Cost monitoring and limits

Integration with Applications

• Web and mobile app integration
• Backend service patterns
• CMS and DAM integration
• API gateway and security patterns

Testing & Quality Assurance

• Evaluating multimodal outputs
• Consistency and brand checks
• Automated QA strategies
• Human review workflows

Monitoring & Observability

• Tracking multimodal requests
• Logging and tracing
• Drift and quality monitoring
• Alerting and troubleshooting

Hands-on Exercises

Summary and Conclusion