L7. Machine Learning and AI: Vertex AI, AutoML, and Pre-built APIs
Video generating
Check back soon for the video lesson on Machine Learning and AI: Vertex AI, AutoML, and Pre-built APIs
Google Cloud offers ML capabilities for every skill level. The Digital Leader exam tests Vertex AI, pre-built AI APIs, AutoML, and how AI enables business transformation.
Google's AI Philosophy
Google has been an AI-first company since 2016. Google Cloud provides the same AI infrastructure used by Google Search, Gmail, and Google Maps.
Three levels of ML sophistication on Google Cloud:
- Pre-built AI APIs - Use Google's pre-trained models via API (no ML expertise needed)
- AutoML - Train custom models on your own data without coding (low-code)
- Vertex AI Custom Training - Full ML platform for data scientists building custom models
Pre-built AI APIs
Ready-to-use AI capabilities accessed via REST API:
| API | Capability |
|---|---|
| Vision AI | Image labeling, object detection, OCR, face detection, explicit content detection |
| Video AI | Object tracking, activity recognition, transcript extraction |
| Natural Language API | Sentiment analysis, entity extraction, classification, syntax analysis |
| Translation API | 100+ language translation |
| Speech-to-Text | Audio transcription, speaker diarization |
| Text-to-Speech | Synthesize natural-sounding speech |
| Document AI | Extract structured data from documents (invoices, contracts, forms) |
AutoML
AutoML lets you train custom ML models on your own labeled data with no ML expertise or coding. AutoML products:
- AutoML Vision: custom image classifiers and object detectors
- AutoML Natural Language: custom text classifiers
- AutoML Tables: structured data prediction (classification and regression)
- AutoML Translation: custom translation models for domain-specific vocabulary
Vertex AI
Vertex AI is Google's unified machine learning platform for the full ML lifecycle: data preparation, training, evaluation, deployment, and monitoring. Key capabilities:
- Vertex AI Workbench: managed Jupyter notebooks
- AutoML (integrated): train without code
- Custom training: Bring your own code (TensorFlow, PyTorch, scikit-learn)
- Model Registry: manage and version models
- Model monitoring: detect data drift after deployment
- Vertex AI Studio: prompt engineering and fine-tuning for generative AI
Generative AI on Google Cloud
- Gemini: Google's multimodal foundation model (text, images, video, audio)
- Vertex AI Gemini API: access Gemini models for application development
- Agent Builder: build conversational AI agents and chatbots
- ✓Three ML levels: Pre-built APIs (no expertise), AutoML (custom models, low-code), Vertex AI (full platform)
- ✓Vision AI, Natural Language API, Translation, Speech-to-Text are pre-built APIs requiring no ML training
- ✓AutoML trains custom models on your own labeled data without coding; 8 AutoML products available
- ✓Vertex AI is the unified ML platform for the full lifecycle from data prep to model monitoring
- ✓Gemini is Google's multimodal foundation model; Vertex AI Gemini API provides access for developers
1. A company wants to add the ability to extract text from scanned invoice images without building or training any machine learning models. Which Google Cloud service should they use?
2. A data scientist needs to train a custom natural language classification model using their company's proprietary dataset, tune hyperparameters, and deploy the model for inference. Which Google Cloud service provides the full ML lifecycle platform?
Recommended: Pluralsight
Reinforce these lessons with Pluralsight's Google Cloud paths: structured video courses, GCP console labs, and practice exams for the Digital Leader certification.