Patents by Inventor Roli Agrawal

Roli Agrawal has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Learning-augmented application deployment pipeline
    Patent number: 12248771
    Abstract: A method includes providing a neural network with metrics obtained from an execution of an application in a test environment to determine rule-related weights, scaling rule results with the rule-related weights to determine scaled rule results. The method also includes re-training the neural network with the rule results of the application, an indication that the executed application is selected for deployment in the production environment, and rule results of other applications in the test environment in response to a determination that the scaled rule results fail a threshold but that the application is selected for deployment in a production environment. The method also includes providing the re-trained neural network with the rule results to generate updated rule-related weights and scaling the rule results by the updated rule-related weights to determine updated scaled rule results.
    Type: Grant
    Filed: May 9, 2022
    Date of Patent: March 11, 2025
    Assignee: Capital One Services, LLC
    Inventors: Roli Agrawal, Bada Kim, Varun Nalamati, Laxmi Kadariya, Patrick Tirtapraja, Nicholas Sorkin, Frank Huang, Taylor Gaskins, Hasan Alatrakchi
  • LEARNING-AUGMENTED APPLICATION DEPLOYMENT PIPELINE
    Publication number: 20230359449
    Abstract: A method includes providing a neural network with metrics obtained from an execution of an application in a test environment to determine rule-related weights, scaling rule results with the rule-related weights to determine scaled rule results. The method also includes re-training the neural network with the rule results of the application, an indication that the executed application is selected for deployment in the production environment, and rule results of other applications in the test environment in response to a determination that the scaled rule results fail a threshold but that the application is selected for deployment in a production environment. The method also includes providing the re-trained neural network with the rule results to generate updated rule-related weights and scaling the rule results by the updated rule-related weights to determine updated scaled rule results.
    Type: Application
    Filed: May 9, 2022
    Publication date: November 9, 2023
    Applicant: Capital One Services, LLC
    Inventors: Roli Agrawal, Bada Kim, Varun Nalamati, Laxmi Kadariya, Patrick Tirtapraja, Nicholas Sorkin, Frank Huang, Taylor Gaskins, Hasan Alatrakchi