Bedrock overview

Deploy production-ready pipelines and models

Bedrock is a machine learning platform designed for data science teams. It enables you to unleash the power of ML into your core products, by making it dead easy to convert algorithms into production-grade micro-services.

What can you do in Bedrock?

These are the primary components of the machine learning workflow that you can execute in Bedrock:

Workflow Component

Description

Input

Output

Training Pipeline

Trains a machine learning model

Training Code

Model version

Batch Scoring Pipeline

Conducts batch inference on a set of input data and writes to a database

Model version

Entry in database

Serve

Creates model servers deployed behind an endpoint that allows real-time inference

Model version

Model Endpoint

‚Äč

How Bedrock is deployed in your infrastructure?

Bedrock is a managed platform-as-a-service. It orchestrates machine learning infrastructure, jobs and services in your cloud infrastructure. The images shown below explains how Bedrock APIs touch customer infrastructure.