报告详情
Energy-Efficient Cloud–Edge MLOps for Carbon-Aware AI Model Deployment
编号：138
访问权限：仅限参会人
更新：2026-11-10 16:54:12 浏览：10次
Online
报告开始：2026年11月21日 10:45（Asia/Kolkata）
报告时间：15min
所在会场：[S6] Cloud, Edge and Distributed Computing Systems [S6-1] Cloud, Edge and Distributed Computing Systems
暂无文件
摘要
Machine learning (ML) systems are increasingly deployed across distributed cloud and edge environments to support real-time, scalable, and intelligent applications. However, the growing computational demands of model deployment and inference have raised important concerns regarding energy consumption and associated carbon emissions. This study explores an energy-efficient cloud–edge MLOps framework for carbon-aware AI model deployment, focusing on how deployment decisions, workload placement, hardware selection, resource allocation, and inference configurations influence environmental sustainability. The proposed approach considers dynamic workload distribution between cloud and edge resources while incorporating energy efficiency and carbon intensity into deployment and scaling decisions.
The study examines optimization strategies including model compression, quantization, hardware-aware deployment, adaptive batch sizing, intelligent workload scheduling, autoscaling, and carbon-aware resource orchestration. By selecting suitable execution environments and dynamically allocating inference workloads according to computational demand, resource availability, latency requirements, and carbon intensity, the framework aims to reduce energy consumption without significantly compromising model accuracy or service quality. The research further highlights the role of MLOps automation, monitoring, and lifecycle management in continuously evaluating energy usage and carbon emissions during model serving. The proposed cloud–edge deployment approach provides a structured foundation for developing sustainable AI infrastructures that balance predictive performance, operational efficiency, latency, scalability, and environmental impact. This work contributes toward greener MLOps practices and supports the development of carbon-aware AI systems for large-scale enterprise and edge computing applications.
The study examines optimization strategies including model compression, quantization, hardware-aware deployment, adaptive batch sizing, intelligent workload scheduling, autoscaling, and carbon-aware resource orchestration. By selecting suitable execution environments and dynamically allocating inference workloads according to computational demand, resource availability, latency requirements, and carbon intensity, the framework aims to reduce energy consumption without significantly compromising model accuracy or service quality. The research further highlights the role of MLOps automation, monitoring, and lifecycle management in continuously evaluating energy usage and carbon emissions during model serving. The proposed cloud–edge deployment approach provides a structured foundation for developing sustainable AI infrastructures that balance predictive performance, operational efficiency, latency, scalability, and environmental impact. This work contributes toward greener MLOps practices and supports the development of carbon-aware AI systems for large-scale enterprise and edge computing applications.
关键词
Green MLOps, Carbon-Aware Model Deployment, Energy-Efficient Machine Learning
报告人
Sai Prasad Veluru
Software Engineer Software Engineer； JP Morgan Chase稿件作者
全部评论
重要日期
-
会议日期
11月20日
2026
至11月22日
2026
-
10月31日 2026
初稿截稿日期
-
11月07日 2026
提前注册日期
-
11月22日 2026
注册截止日期
主办单位
United Societies of Science
承办单位
Pondicherry University
协办单位
IEEE Section
IEEE Madras Section
IEEE Madras Section
联系方式
- ICon INDIA 2026
- ic******@usssociety.org
发表评论