MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202611075244 A) filed by Dr. Amit Kumar Dubey on June 17, 2026, for System And Method For Predictive Cloud Resource Allocation During Ai Model Execution.
Inventors include Dr. Vikas Sethi; Dr. Rashmi Sindhu; Dr. Deepti Dhingra; Kiran; and Sweety.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: The present invention relates to a system and method for adaptive cloud resource allocation for artificial intelligence model execution using predictive workload engines. The system comprises a cloud telemetry acquisition module, an artificial intelligence workload profiling module, a predictive workload engine, a resource mapping controller, an orchestration controller, a cloud resource pool, and an execution monitoring module. The telemetry acquisition module collects real-time parameters including processor utilization, accelerator load, memory usage, network bandwidth, storage access rate, execution latency, and queue depth. The workload profiling module identifies model type, execution stage, batch size, input data volume, and execution priority. The predictive workload engine estimates future resource demand before resource saturation occurs. The resource mapping controller generates allocation instructions, and the orchestration controller provisions, migrates, modifies, or releases cloud resources accordingly. The system improves cloud infrastructure performance by reducing latency, preventing over-provisioning, improving accelerator utilization, reducing execution failures, and enabling reliable AI model execution. (Accompanied Figure No. 1-2)
Disclaimer: Curated by HT Syndication.
