Are you looking for a single development environment for the entire data science workflow? Do you need a notebook-based environment to query and explore data? Are you planning to develop and train a model, and run your code as part of a pipeline?
Then look no further than Vertex AI. The notebook-based environment of Vertex AI Workbench allows you to run your code as part of a pipeline, develop and train a model, and query and explore data.
Vertex AI assists you with data preparation. You can use Vertex AI Data Labelling to annotate high-quality training data and improve prediction accuracy by ingesting data from BigQuery and Cloud Storage.
To serve, share, and reuse ML features, use Vertex AI Feature Store, a fully managed rich feature repository. Track, analyze, and discover ML experiments with Vertex AI Experiments for faster model selection. To visualize ML experiments, use Vertex AI TensorBoard. Vertex AI Pipelines can help you simplify the MLOps process by streamlining the creation and execution of ML pipelines.
Create cutting-edge ML models without writing code by using AutoML to determine the best model architecture for your image, tabular, text, or video prediction task, or by using Notebooks to create custom models. Vertex AI Training provides fully managed training services, while Vertex AI Vizier provides hyperparameter optimization for maximum predictive accuracy.
Vertex Explainable AI provides detailed model evaluation metrics and features attributions. Vertex Explainable AI indicates the importance of each input feature to your prediction. Available right away in AutoML Forecasting, and Vertex AI Prediction.
Vertex AI assists you in moving from notebook code to a cloud-deployed model. Vertex AI has every tool you need, from data to training, batch or online predictions, tuning, scaling, and experiment tracking.
Vertex AI Prediction makes it simple to deploy models for online serving via HTTP or batch prediction for bulk scoring. You can deploy custom models built on any framework to Vertex AI Prediction, including TensorFlow, PyTorch, scikit, or XGB, with built-in tooling to track your models’ performance.
For models deployed in the Vertex AI Prediction service, continuous monitoring allows for easy and proactive monitoring of model performance over time. Continuous monitoring continuously monitors signals for the predictive performance of your model and alerts you when they deviate, diagnoses the cause of the deviation, and triggers model-retraining pipelines or collects relevant training data.
Vertex ML Metadata automates the tracking of inputs and outputs to all components in Vertex Pipelines for artefact, lineage, and execution tracking, making it easier to audit and govern your ML workflow. Custom metadata can be tracked directly from your code and queried using a Python SDK.
Thus, Data Scientists can use Vertex AI to train models without writing code, accelerate models to production, and confidently manage machine learning models.