Amazon Web Services’ (AWS) artificial intelligence (AI) portfolio is a collection of machine learning (ML) and AI solutions for the data science market.
Seattle-based AWS has about 50,000 employees, working on cloud-based solutions such as AI in various regions around the globe.
AWS reported $106.3 million in artificial intelligence revenue in 2019, according to an 2020 report by IDC.
See below to learn about the broad set of AWS’ ML and AI offerings:
AWS AI Portfolio
Amazon SageMaker offers fast methods for training deep learning models and data sets, using data parallelism and model parallelism.
- Can be implemented with a few lines of code
- Uses graph-partitioning algorithms to determine the best model-splitting approach
- Optimizes distributed training tasks to fully utilize infrastructure resources
Amazon SageMaker Model Monitor
SageMaker Model Monitor is a fully managed service that continuously monitors machine learning models during their production phase.
- Detects data deviations
- Sends out early alerts
- Built-in analysis tools
- Integrates with various SageMaker products
Amazon SageMaker AutoPilot
SageMaker Autopilot eliminates a portion of the heavy lifting that goes into building ML models and automatically builds, trains, and tunes ML models based on available data.
- Automatically fills in missing data
- Automatically selects from a collection of ML models
- Features priority-based progress reports
Amazon SageMaker Ground Truth
SageMaker Ground Truth is Amazon’s fully managed data-labeling service. It allows users to train ML models using accurately labeled and semi-labeled objects and data points.
SageMaker Ground Truth supports various data types, including 2D images, 3D models and point clouds, videos, images, and text.
- Reduces costs by up to 70%
- Intuitive user interface
- Time-efficient with worker selection
- 2D and 3D object treatment
Amazon SageMaker JumpStart
SageMaker JumpStart provides a set of solutions that kickstarts ML model development. It supports one-click deployment for the most popular open-source ML models.
There are numerous ways SageMaker JumpStart can be used, such as:
- Fraud detection
- Predictive maintenance
- Computer vision
- Demand forecasting
- Personalized recommendations
Amazon SageMaker Data Wrangler
SageMaker Data Wrangler is a cloud solution that reduces the time it takes to aggregate and prepare data for training ML models from “weeks to minutes.”
- Contains over 300 built-in data transformations
- Quick previews with data visualization templates
- Diagnoses and fixes ML data issues
- Automates data preparations workflows
Amazon SageMaker Feature Store
SageMaker Feature Store is a fully managed repository to store, update, retrieve, and share ML model features.
It offers a unified storefront for features during real-time training and keeps services updated as new data gets generated.
- Multi-source data ingestion
- Search- and discovery-based indexing system
- Enforces feature standardization
Amazon SageMaker Clarify
SageMaker Clarify provides model and training data visibility to machine learning developers to identify and minimize bias.
- Feature importance graphs
- Monitors ML models for changes in behavior
- Detects data imbalances
- Continuously checks trained models for biases
Amazon SageMaker Debugger
SageMaker Debugger continuously monitors how system resources are utilized and collects data to optimize ML models in real-time during productivity loss. It monitors CPUs, GPUs, and network and memory usage.
- Automatic detection, alerts, and analysis
- Built-in analysis tools
- Supports a broad range of ML algorithms and frameworks
Amazon SageMaker Studio
SageMaker Studio is a centralized web-based visual interface that allows users to perform all primary ML development procedures. It offers complete access, control, and visibility in each step required to build and deploy ML models.
- Built-in elastic and shareable Jupyter Notebooks
- Scalable data preparations, exploration, and visualization using Scala, SQL, or Python
- Over 150 open source ML models
- Over 15 pre-built solutions for model building
- Supports a variety of ML and AI frameworks and libraries
Amazon SageMaker Partners is AWS’ partnership program for ML experts, where they work on accelerating the process of solving complex AI business problems using ML solutions.
They offer two types of SageMaker partnerships: The Consulting Partners program offers consulting services for Amazon SageMaker. The ISV Partners program offers exclusively vetted and validated solutions that have demonstrated technical proficiency and customer success.
AWS AI use case
One of AWS’ AI clients is Zendesk.
With nearly 100,000 paying customers in over 150 countries and territories using Zendesk products, they needed a way to scale up operations without sacrificing quality.
“Amazon SageMaker will lower our costs and increase velocity for our use of machine learning,” says Davis Bernstein, director of strategic technology at Zendesk.
“With Amazon SageMaker, we can transition from our existing self-managed … deployment to a fully managed service.
User reviews of AWS AI
AWS’ AI portfolio and SageMaker line score high ratings on a number of third-party review websites.
G2: 4.2 out of 5
TrustRadius: 8.1 out of 10
Gartner Peer Insights: 4.7 out of 5
Industry recognition of AWS AI
Amazon SageMaker was named the Outright Leader in Enterprise MLOps Platforms in 2021 by Omdia, after launching in 2017.
“Across almost every measure, the company significantly outscored its rivals, delivering consistent value across the entire ML life cycle,” said Bradley Shimmin, chief analyst at Omdia.
AWS in the AI market
AWS holds the fifth largest share of the AI software market (3.1% in 2019), according to a 2020 report by IDC.
In comparison, IBM holds the largest share of the AI software market in the report (8.8%), and SAS ranks third (4.4%).
The AI software market was worth an estimated $3.5 billion in 2019, IDC says.