Manish Mitruka

Title: Manish Mitruka: Innovating in AI-Driven MLOps

Introduction

Manish Mitruka is an inventor based in Pune, India, whose work focuses on machine learning operations and cloud-based systems. He is associated with Kyndryl, Inc. His patent describes an approach to managing machine learning models for cost forecasting in a multi-tenant cloud environment.

Latest Patents

Mitruka is named on 1 patent: “AI Customized MLOps Framework in a Multi-Tenant Cloud Environment.” The invention describes a computer-implemented system, method, and program product for managing machine learning operations for cost forecasting models.

The system receives a tenant’s request for a subscription machine learning model and uses the tenant’s configuration profile to select a suitable model from a registry. It deploys the selected model and monitors its usage at a predetermined refresh frequency. The system also evaluates whether the operating model exceeds a predetermined accuracy threshold and, in response, automatically replaces it with a second subscription model from the registry.

Career Highlights

Mitruka’s patent highlights an approach to tailoring model selection and deployment to individual tenant configurations. Its automated monitoring and model replacement process is designed to help maintain accuracy as machine learning models operate in a shared cloud environment.

Collaborations

Mitruka is associated with Kyndryl, Inc. His listed coworkers include Omar Odibat and Umar Mohamed Iyoob Umar.

Conclusion

Manish Mitruka’s patented work addresses the selection, monitoring, and replacement of machine learning models for cost forecasting, with a focus on tenant-specific needs in multi-tenant cloud environments.