GCP-GAIL: Exam Objectives
Key Resource Links
| Resource | Description |
|---|---|
| Official Exam Guide | Lists the four domains: Fundamentals, Offerings, Techniques, and Business Strategy |
| Official Learning Path | Free Skills Boost course series (~8 hours) specifically for the Leader exam |
| Vertex AI Documentation | Product details - difference between Vertex AI Studio and Google AI Studio |
Exam Weighting
| Domain | Weight | Focus Areas |
|---|---|---|
| Domain 1: Vertex AI Foundation | ~30% | Model Garden, Studio, API integration, model selection |
| Domain 2: Prompt Engineering | ~25% | Few-shot, CoT, parameters (Temperature, Top-K, Top-P) |
| Domain 3: Data & Customization | ~25% | RAG, Vector Search, Fine-tuning, Grounding |
| Domain 4: Responsible AI & Ops | ~20% | Safety filters, evaluation, monitoring, bias mitigation |
In-Scope Services & Technologies
Gemini Model Family
- Gemini Ultra: Most capable, complex reasoning and coding
- Gemini Pro: General purpose, balanced performance
- Gemini Flash: Speed optimized, high-throughput
- Gemini Nano: On-device, mobile/edge deployment
Generative AI Studio
- Language: Text generation, chat, code (Codey)
- Vision: Image generation (Imagen), visual Q&A
- Speech: Speech-to-Text (Chirp), Text-to-Speech
Model Garden
- First-party: Gemini, Imagen, Codey, Chirp
- Open-source: Llama, Mistral, Gemma
- Third-party: Claude (Anthropic)
Model Customization
- Prompt Design: Zero-shot, few-shot, chain of thought
- Fine-tuning (SFT): Supervised fine-tuning with JSONL datasets
- Distillation: Creating smaller models from larger ones
- Grounding: Google Search, custom data sources
RAG & Data Infrastructure
- Vertex AI Vector Search: High-scale embedding search (formerly Matching Engine)
- Embeddings API: Text/image to vector conversion
- Document AI: Document parsing and extraction
- BigQuery ML: In-warehouse inference
Operations & Evaluation
- AutoSxS: Side-by-side model comparison
- Model Monitoring: Drift detection, performance tracking
- Safety Filters: Configurable content filtering thresholds
- Endpoints: Model deployment and serving
Study Progress
GCP-GAIL Study Progress
0/50%
💾 Progress is saved in your browser's local storage. Clearing your browser data will reset your progress.