Machine Learning with Google Cloud Platform
summary
Overview of concrete Google Cloud ML capabilities: managed Vertex AI for notebooks, training (Docker containers), model registry and endpoints; a feature store that ingests from BigQuery in batch or from Pub/Sub with Apache Beam/Dataflow for real‑time features; MLOps pipelines (TFX/Kubeflow) with lineage, monitoring and automated retrain; explainability via What‑If/LIME/SHAP and metadata tracking; NAS and Vizier hyperparameter tuning using GPUs/TPUs. Example: fraud detector on millions of BigQuery transactions moves from 10–20 prototype features to 40 and in some cases 80–200 in production.
AI-written summary · about
chapters (9)
- 0:00ML on Google Cloud intro
- 3:59For data scientists
- 10:36Feature store for ML
- 15:34Vertex AI pipelines
- 17:18MLOps automation
- 21:53AI model explainability
- 24:43Vertex AI advanced algorithms
- 27:57Google Cloud AI recap
- 30:04Q&A
related talks
29mML Ops on Google Cloud Platform
23mBuilding Scalable, Observable MLOps Systems on Google Cloud
59mMachine Intelligence at Google Scale
31mProductionizing ML with ML Ops and Cloud AI
03mML Ops on Google Cloud
44mBuilding smarter solutions with no expertise in machine learning
27mKeynote - Google Cloud, Machine Learning
42mMachine Learning APIs, from Training [...]