Amazon Technologies, Inc.

Tanya Bansal

Rajesh Viswanathan

Veeraraghavan Vijayaraj

William Trotter Shelton

Prashant Prahlad

Adam Daniel Everett Wright

Benjamin John Ullom

Stefano Stefani

Leo Parker Dirac

Vladimir Zhukov

Laurence Louis Eric Rouesnel

Orchid Majumder

Ravikumar Anantakrishnan Venkateswar

Patricia Grao Gil

Philip Gautier

Fan Li

Piali Das

Saurabh Gupta

William Frederick Hingle Kruse

Christopher Whitaker

Mukesh Karki

Zohar Karnin

Lakshmi Naarayanan Ramakrishnan

Ravikanth Repaka

Rodolfo Flores Hinojosa

Xingyuan Wang

Alex Tang

Vidhi Kastuar

Techniques for automated machine learning (ML) pipeline exploration and deployment are described. An automated ML pipeline generation system allows users to easily construct optimized ML pipelines by providing a dataset, identifying a target column in the dataset, and providing an exploration budget. Multiple candidate ML pipelines can be identified and evaluated through an exploration process, and a best ML pipeline can be provided to the requesting user or deployed for production inference. Users can configure, monitor, and adapt the exploration at multiple points in time throughout.

Automated machine learning pipeline exploration and deployment

Techniques and technologies for providing a fully managed datastore for clients to securely store, discover, retrieve, remove, and share curated data, or features, to develop machine learning (ML) models in an efficient manner. The feature store service may provide clients with the ability to create and store feature groups that include features and associated metadata providing clients with a quick understanding of features so that they may determine which features are suitable for training ML models and/or use with ML models. The feature store service may provide first a data store configured to store the most recent values associated with a feature group, such that client can access the features and utilize ML models to make real-time predictions with low latency and high throughput, and a second datastore configured to store historical values associated with a feature group, such that a client can utilize the features to train ML models.

Fully managed repository to create, version, and share curated data for machine learning development

A configuration management service provides data identifying its subscribers to a secure sharing service that executes in an account that has a higher security level than a service account used to provide the configuration management service. The secure sharing service securely determines whether each subscriber has authorized producer services to share resource configuration data with the configuration management service. If a subscriber has authorized such sharing, information identifying the subscriber can be stored in a location accessible to the producer services. If a subscriber has not authorized such sharing, the secure sharing service will not make the subscriber's information available to the producer services. The producer services can use the subscriber data to provide resource configuration data to the configuration management service only for those subscribers that subscribe to both the configuration management service and to the producer services.

Secure and efficient cross-service sharing of subscriber data

Configurations of computing resources established in a network-based computing service can be tested to assess impacts of proposed changes to the configuration states of the computing resources or to policies governing the configurations of such computing resources. Test environments can be established to simulate an operational computing environment under study. Changes to the configuration of one or more computing resources can be introduced to the test environment, and impacts to the test environment can be assessed. Changes to configuration policy can also be introduced and subsequent impacts can be observed.

Resource configuration testing service

A configuration management system may receive information defining a threshold range of deviation from a baseline configuration of at least one computing resource. The system may acquire periodic records of configuration information, from which it may be determined if the current configuration at least one computing resource has deviation from the baseline outside of the tolerated threshold. A record indicative of a prior configuration within a threshold configuration deviation may be identified. A template may be formed based on the record. The at least one computing resource may be reconfigured, based on the template, to be within the threshold configuration deviation.

Computing resources checkpoint restoration

A configuration history service for web-based computing resources is implemented by capturing and storing a historical record of each configuration state for each computing resource in a subscriber's account. Files of the configuration history store information about a computing resource type offered by the web service. The file comprises records for each particular resource of that type that is or was actively configured in the subscriber's account during the period that the configuration history service is active in the account. Each record lists a particular configuration state, at a particular point in time, for a particular computing resource. The configuration history of a computing resource can be displayed as a timeline by which a subscriber can access information describing the configuration state of the particular computing resource at any point in time.

Resource configuration history service

Configurations of computing resources established on a web-based computing service can be monitored, managed and controlled. According to one embodiment, a method may include discovering a set of computing resources being used by an enterprise. The disclosed resource configuration service can deliver a set of resource-specific data that uniquely identifies each computing resource and describes each resource's configuration state. The resource configuration service monitors the computing environment and detects changes to the configurations of computing resources. The resource configuration service can deliver a stream of notifications when computing resources are created, updated, deleted or otherwise reconfigured. The notifications can be used by, for example, software tools or system administrators.

Resource configuration service

Configurations of computing resources established on a web-based computing service can be monitored, managed and controlled. According to an embodiment, a configuration compliance service is implemented. Policy rules governing attributes of configuration states for computing resources in a subscriber's account are defined. The computing resources in the account are monitored for changes to configuration states. In response to detection of a change to a configuration state that violates a rule, a predefined action is taken. The predefined action can include sending a notification to the subscriber's account, reconfiguring the configuration state of the computing resource, and deactivating the computing resource.

Resource configuration compliance service

Growing community of inventors

Redmond, WA, United States of America

Tanya Bansal