Abstract: Techniques are disclosed relating to backing up skip list data structures to facilitate a subsequent recovery. In various embodiments, a computing system creates a checkpoint of a skip list including a plurality of key-value records that include pointers to others of the plurality of key-value records. Creating the checkpoint includes scanning the skip list to identify ones of the plurality of key-value records that are relevant to the checkpoint and storing the identified key-value records in a storage such that the identified key-value records include pointers modified to exclude ones of the plurality of key-value records that are not relevant to the checkpoint. The computing system can then recover the skip list based on the created checkpoint.

Abstract: Systems, devices, and techniques are disclosed for security configuration evaluation. A binary representation of a reference security configuration for an application may be generated. The binary representation of the reference security configuration for the application may be hashed to generate a reference hash for the application. Data for an instance security configuration for an instance of the application may be received. A binary representation of the instance security configuration may be generated from the received data for the instance security configuration. The binary representation of the instance security configuration may be hashed to generate an instance hash. The computing device may determine the distance between the reference hash and the instance hash. The instance security configuration may be determined to be secure if the distance is not greater than a threshold.

Abstract: An apparatus is provided for outputting suggested query items by a group-based communication platform. The apparatus may include at least one memory and at least one processor configured to detect group-based communication platform query data stored in a memory of the apparatus. The processor is also configured to rank suggested query items based on the group-based communication platform query data. The processor is further configured to detect non-keystroke engagement of a search input interface of the group-based communication platform. The processor is further configured to determine a first subset of the suggested query items based on the ranking of the suggested query items and automatically launch a query suggestion interface including a first subset of the suggested query items in response to detecting the non-keystroke engagement of the search input interface. Corresponding methods and computer program products are also provided.

Abstract: Systems, devices, and techniques are disclosed for multi-threaded dynamic queries on an unpartitioned database. An instruction to perform a database operation on rows of a database may be received. Processing threads may be generated by generating three random numbers. A starting point of a processing thread may be set to a first random number, an offset may be set to a second random number, and an order may be based on a third random number. The processing threads may each generate a dynamic query. A dynamic query generated by a processing thread may include an instruction to perform the database operation on rows of the database based on the offset, starting point, and order for the processing thread. Each dynamic query may be run against the database to perform the database operation.

Abstract: Some implementations of the disclosed systems, apparatus, methods and computer program products may provide for chatbots configured to perform tasks requiring end user identification on behalf of users. Such a chatbot may be authenticated through tokens with custom claims. The custom claims may include identifying or authenticating tokens received by the chatbot or server system and the chatbot may create and/or provide such tokens for authentication. The custom claim may be configured to provide user identifying data, allowing for the chatbot to be provided with end user credentials. Accordingly, chatbots may be utilized to perform sensitive tasks that require user credentials while continuing to provide security for users.

Abstract: System and methods are described for implementing controlled updates of containers in a distributed application deployment environment. In one implementation, a method comprises: receiving a request to update a target container of a plurality of containers within a deployed computing unit; identifying an updated configuration object to be applied to the target container based on the request; receiving or computing a health indicator representative of performance of the deployed computing unit or other containers to which the updated configuration object has been applied; and assigning the updated configuration object to the target container based on the health indicator.

Abstract: Disclosed are examples of systems, apparatuses, methods, and computer program products for dynamic traffic throttling. A server system can receive, via an edge worker associated with a content delivery network (CDN), a first set of requests to access a first site. The server system can determine that traffic to the first site is to be throttled. The server system can determine a throttling rate. The server system can transmit instructions to the edge worker, the instructions configured to cause the edge worker to direct at least a portion of a second set of requests to access the first site to a waiting room site prior to being directed to the first site, the direction of the at least the portion of the second set of requests to the waiting room site being subject to the throttling rate.

Abstract: In the present disclosure, systems and methods are described for allowing a non-code user to create to transform a database in an ETL pipeline. Specifically, as disclosed herein, a user can take a database and receive a ruleset to apply to the database in an ETL pipeline. The data linting system may take the database and extract a schema and a data sample from it. Further, the data linting system may use the schema and data sample to create two rulesets. With these rulesets, the data linting system combines them to create a final ruleset which may be validated using the data sample. The data linting system then sends the final ruleset and the validation report to the user. With this system, the user only needs to give it a database and will receive a ruleset that is able to be immediately used in an ETL pipeline.

Abstract: Techniques are disclosed relating to multi-tenant database systems with underlying key/value storage. In some embodiments, a computing system generates first logical table information for a first portion of a multi-tenant dataset corresponding to a first tenant and second logical table information for a second portion of the dataset corresponding to a second tenant. The system may store the first and second logical tables using non-relational key/value data structures in one or more storage elements, including: storing entries from the first logical table across a first set of multiple key/value store buckets and storing entries from the second logical table across a second set of multiple key/value store buckets. Disclosed techniques may reduce performance issues due to hotspots while allowing efficient data access for a given tenant by limiting the number of buckets per tenant, relative to traditional techniques.

Abstract: Various techniques and mechanisms for sharing remote resources among a trusted group are disclosed. A credential management agent utilizes a resource credential for a first user to access a secure resource corresponding to the first user for a second user by at least validating a second user and validating a consent of the first user to allow the second user to access the secure resource using the resource credential for the first user. The secure resource resides on a remote server system accessible via one or more application program interfaces (APIs). A platform management agent provides an interface for shared resource-agnostic credential sharing. The platform management agent validates credentials for the second user as belonging to a trusted group and forwards a request for access to the secure resource for the second user to the credential management agent.

Abstract: Embodiments described herein provide visual-and-language (V+L) systems and methods for learning vision and language representations. Specifically, a method may comprise receiving a training dataset comprising a plurality of image samples and a plurality of text samples; encoding the plurality of image samples into a plurality of encoded image samples and the plurality of text samples into a plurality of encoded text samples; computing a first loss objective based on the plurality of encoded image samples and the plurality of encoded text samples; encoding a first subset of the plurality of encoded image samples and a second subset of the plurality of encoded text samples into a plurality of encoded image-text samples; computing a second loss objective based on the plurality of encoded image-text samples; and updating the V+L model based at least in part on the first loss objective and the second loss objective.

Abstract: Systems and methods may include obtaining data identifying a plurality of cancelled health-related appointments for a plurality of patient records stored in a database, the cancelled health-related appointments associated with a shared health event; associating each of the plurality of patient records with a risk category from a plurality of risk categories ranging from a lowest risk to a highest risk; communicating with a client computing device associated with a patient having a cancelled health-related appointment to schedule a return appointment based at least on a risk category, the communicating including receiving current health condition information of the patient; based on the current health condition information of the patient showing symptoms consistent with symptoms associated with the shared health event, identifying the patient as a potential infected patient; and performing contact tracing based at least on people that the patient has contact with while experiencing the symptoms.

Abstract: A method receives an input file. The input file includes a plurality of entries where different entries are associated with a different characteristic. A selection of a guide for an entry is determined where a guide includes guidelines for formatting content in an entry in the plurality of entries. The method performs analyzing an image from the entry to generate image variations. Then, the method analyzes the image and the image variations to place text items from the entry in the image and the image variations where the text items in the image and the image variations are formatted based on the guidelines in the guide for the entry. A plurality of design variants is generated for the entry based on the image and the image variations that include text items.

Abstract: The disclosed techniques automatically ingest new documents and store data extracted from the documents in a database for conversion into a different format. The disclosed techniques identify, via a backend API, newly released documents that include data for users and, based on the identifying, automatically ingest, via an ingestion call executed made by the backend API, the newly released documents. The disclosed techniques extract, using a computer vision model trained on different types of documents, a data from the newly released documents, where the extracting includes identifying locations within the documents from which to extract data. The disclosed techniques store the extracted data in the database storing data extracted from previously ingested documents for users in a text-based object format and convert, using a machine learning model trained on a plurality of metatags, data corresponding to a given user from the text-based object format to a queryable file format.

Abstract: Systems, devices, and techniques are disclosed for cryptographic key migration. A tenant host may determine a first Key Management Service (KMS) indicated as storing a cryptographic key associated with the tenant host from a new KMS mapping. The tenant host may send a request for the cryptographic key associated with the tenant host to the first KMS. The tenant host may receive an indication from the first KMS that the first KMS does not store the cryptographic key. The tenant host may determine a second KMS indicated as storing the cryptographic key associated with the tenant host from an original KMS mapping. The tenant host may receive the cryptographic key associated with the tenant host from the second KMS. The tenant host may send a request to the second KMS that the cryptographic key associated with the tenant host be replicated from the second KMS to the first KMS.

Abstract: Embodiments described herein adopts a k nearest neighbor (kNN) mechanism over a model's hidden representations to identify training examples closest to a given test example. Specifically, a training set of sequences and a test sequence are received, each of which is mapped to a respective hidden representation vector using a base model. A set of indices for each sequence index that minimizes a distance between the respective hidden state vector and a test hidden state vector is then determined A weighted k-nearest neighbor probability score can then be computed from the set of indices to generate a probability distribution over labels for the test sequence.

Abstract: Embodiments described herein provide a multi-modal search-and-summarize tool for message platforms. Specifically, the multi-modal search-and-summarize tool may monitor conversational content of different formats, e.g., text, image, video, etc., and use multi-modal summarization models to generate a summary of the conversation channel. The summarization may be conducted via a search-and-summarize process in response to a specific user query, e.g., a user may enter “what did John and Josh say about the presentation tomorrow?” The multi-modal summarization model would first search for relevant conversation messages between user John and user Josh, identify communication files of different format (e.g., text messages, emojis, multimedia attachments, etc.), and then input the communication files to respective text or image encoders to generate a summary of the communication content.

Abstract: Connections among two different entities are identified by processing a collection of documents to identify one or more documents that contain co-occurring mentions of each of the two different entities. This relationship may be graphically displayed in a user interface with an icon or the like for each of the two entities, along with a graphical link interconnecting the entities. The graphical link can be an active element of the user interface that responds to user interactions by providing access to evidence within the collection of documents that substantiates the connection between the two entities. In one aspect, a search input form field in a user interface may be used to explicitly request documents that substantiate a relationship between two entities. In another aspect, a user may ground entity mentions by explicitly selecting documents that mention an entity of interest.

Abstract: System and method for fine granularity control of data access and usage for across multi-tenant systems. A user makes a request to access a particular set of data from a particular remote data source for a specific purpose. The system authorizes the user to validate whether the user is qualified to make the request. The data source is checked to see if the particular data has been granted access for that particular purpose. A cloud neutral token is created and converted into a cloud specific token upon reaching the remote data source. The cloud specific token is used to create a temporary IAM role and IAM policy with a predetermined time to live. After the time to live expires, the IAM role and IAM policy are deleted.

Abstract: System and method for fine granularity control of data access and usage for across multi-tenant systems. A user makes a request to access a particular set of data from a particular remote data source for a specific purpose. The system authorizes the user to validate whether the user is qualified to make the request. The data source is checked to see if the particular data has been granted access for that particular purpose. A cloud neutral token is created and converted into a cloud specific token upon reaching the remote data source. The cloud specific token is used to create a temporary IAM role and IAM policy with a predetermined time to live. After the time to live expires, the IAM role and IAM policy are deleted.