Deployment & Scaling · Practitioner
Embeddings and Vector Search on Google Cloud
Turn text, images and documents into vectors with Google's embedding models, then serve nearest-neighbour retrieval at scale — including the index cost model that catches people out.
- 1.3 hours lesson
- intermediate level
- Google Cloud AI Track
- Narrated presentation
What you’ll learn
- Text, multimodal and on-device (EmbeddingGemma) embedding models
- Task type parameter — RETRIEVAL_DOCUMENT vs RETRIEVAL_QUERY — and the bug from ignoring it
- Matryoshka dimensionality truncation as a cost/accuracy dial
- Index, index endpoint and deployment as three separate resources
- Namespace and numeric filtering; hybrid dense-plus-sparse search
- Deployed index endpoints bill hourly whether queried or not
- Choosing pgvector, BigQuery or Firestore when scale does not justify Vector Search
Inside this lesson
A narrated, slide-by-slide presentation with 1 prediction exercise, 1 knowledge check, 4 videos. You answer as you go, and the platform checks your work.