Orchestration of AI Services on Telco Cloud Platforms: Dynamic Models, Energy Efficiency and Edge-Cloud Deployment

The Inria Centre at Rennes University is one of Inria's eight centres and has more than thirty research teams. The Inria Centre is a major and recognized player in the field of digital sciences. It is at the heart of a rich R&D and innovation ecosystem: highly innovative PMEs, large industrial groups, competitiveness clusters, research and higher education players, laboratories of excellence, technological research institute, etc.

Funding Context: PEPR NF-MUST

This position is funded within the PEPR 5G and Networks of the Future programme, a national priority research programme (France 2030) co-directed by CEA, CNRS and IMT with a total budget of €65M. The programme aims to position France at the forefront of 5G, 6G and future network technologies across the full value chain.

The present work is part of the NF-MUST project (End-to-End Multi-domain Service Management Architecture of the Networks of the Future), which focuses on automating the provisioning and lifecycle management of multi-domain, multi-stakeholder services over highly heterogeneous and dynamically evolving future network infrastructures. NF-MUST covers end-to-end orchestration of coordination, cooperation and interaction functions to satisfy diverse service requests across multiple sectors, with strong emphasis on resource availability, security, performance and frugality. The project runs from May 2023 to December 2027 and involves partners including CNRS, Inria, CEA-List, Télécom Paris, Télécom SudParis, EURECOM and others.

This research engineer position contributes to the NF-MUST objectives by developing orchestration mechanisms for native AI services deployed over Telco Cloud infrastructure, at the intersection of AI workload management, cloud-native networking and future network architectures.

Scientific Context

5G and pre-6G networks must host heterogeneous intelligent applications — autonomous driving, augmented reality, real-time video analytics, embedded machine learning — whose requirements in terms of latency, energy, and model quality evolve dynamically. Open Telco Cloud platforms, based on Kubernetes, provide a shared infrastructure across operators for hosting cloud-native network functions and edge workloads. A major challenge remains, however: these platforms do not natively handle the specificities of AI workloads — adaptive architectures, distributed training, energy-aware inference scheduling.

This position is part of a research project aimed at designing and validating intelligent orchestration mechanisms for dynamic AI models on Telco Cloud infrastructure, covering the device–edge–cloud continuum. The solutions developed will be designed to be compatible with open standards in the field (Kubernetes/CaaS) and potentially integrable into different Telco Cloud platforms (Sylva, Nephio, or other cloud-native stacks).

General Mission

The research engineer will contribute, according to their profile and the project's priorities, to one or more of the following research tracks:

Track A — System Prototyping and Integration

Track B — Algorithmic Development and Experimentation

Track C — Data Collection and Experimental Validation

Track A — System Prototyping and Integration

• Setup and configuration of a Kubernetes-based Telco Cloud environment (CaaS)

• Deployment and containerisation of AI services (inference models, intelligent network functions) on a heterogeneous testbed platform

• Development of monitoring and profiling tools (latency, energy, accuracy) for dynamic AI workloads

Track B — Algorithmic Development and Experimentation

• Implementation and evaluation of orchestration algorithms (heuristics, optimisation, adaptive policies) for AI model placement across heterogeneous nodes (GPUs, NPUs, edge servers, cloud)

• Design of dynamic model configuration selection policies (early-exit, compression, mixed precision) based on network conditions and available resources

• Comparative evaluation against baselines (static deployment, greedy heuristics) under varying load and network conditions

Track C — Data Collection and Experimental Validation

• Construction of representative evaluation scenarios: video analytics, sensor fusion, intelligent network functions (traffic prediction, anomaly detection)

• Experimental measurement campaigns on the testbed, analysis of energy–latency–quality trade-offs

• Contribution to the production of open-source artefacts (code, datasets, configurations) and to the writing of scientific or technical deliverables

Education: Master's degree or PhD in computer science, networking, distributed systems, or a related field.

Required skills

• Strong programming skills in Python

• Knowledge of machine learning frameworks and familiarity with model training and inference pipelines

• Understanding of distributed systems concepts (scheduling, resource management, containerisation)

• Subsidized meals
• Partial reimbursement of public transport costs
• Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
• Possibility of teleworking (after 6 months of employment) and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities
• Access to vocational training
• Social security coverage

monthly gross salary from 2675 euros according to diploma and experience