Agentic ai For Edge Computing Intelligence H/F

INRIA

Agentic AI for Edge Computing Intelligence
Le descriptif de l'offre ci-dessous est en Anglais
Type de contrat : Convention de stage

Niveau de diplôme exigé : Bac +4 ou équivalent

Fonction : Stagiaire de la recherche

A propos du centre ou de la direction fonctionnelle

The Inria Centre at Rennes University is one of Inria's nine 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.

Contexte et atouts du poste

The successful candidate will join a dynamic and international research team at IMT Atlantique, Nantes (https://stack.inria.fr). As a research intern,you will contribute to the design, implementation and validation of agentic services for data-driven applications across Device-Edge-Cloud computing infrastructures.

Edge intelligence enhances AI capabilities by bringing computation closer to data sources (e.g., IoT, vehicles), improving latency, bandwidth usage, privacy, and responsiveness. Traditional edge AI approaches typically involve static models and centralized orchestration. In contrast, Agentic AI empowers decentralized agents to adapt, reason, and act autonomously based on local context and evolving workloads - an emerging paradigm for truly general edge intelligence.

The successful candidate will be supervised by Daniel Balouek (Inria STACK, IMT Atlantique).

This work will be conducted as part of the QUICK project (\_Collaborative services for Urgent systems across the Edge-Cloud Computing Continuum\_, 2025-2027), funded by the "Etoiles Montantes" regional award.

Mission confiée

The internship will focus on computation offloading, task scheduling, and adaptive resource allocation. Baseline methods (heuristics and classical ML-based approaches) will be implemented first. These will be extended using agentic AI techniques such as reinforcement learning and multi-agent systems, where each edge node acts as an autonomous agent observing its local state and optimizing decisions based on learned policies.

Principales activités

Define, implement, and evaluate agentic behaviors for distributed edge nodes, leveraging scalable experiments on the Grid'5000 testbed and algorithmic insights from Edge-Cloud operations, incorporating:
- Survey Edge Intelligence operations such as computation offloading, data pipelines and on-edge learning to extract key algorithmic techniques and metrics.
- Define agentic AI behaviors (state, actions, reward) suitable for edge resource management;
- Develop agentic policy models using reinforcement learning (RL) or multi-agent reinforcement learning (MARL) to autonomously decide:

- when and where to offload tasks,
- how to allocate resources,
- how to adapt to network variability

4. Evaluate:

- Latency and throughput improvements over baseline strategies.
- Energy/CPU utilization and stability under load.
- Agent adaptability to changing workloads & link conditions

References

Daniel Balouek-Thomert, Ivan Rodero, and Manish Parashar. Harnessing the computing continuum for urgent science. SIGMETRICS Perform. Eval. Rev., 48(2) :41-46, November 2020.

Zhang, Ruichen, Guangyuan Liu, Yinqiu Liu, Changyuan Zhao, Jiacheng Wang, Yunting Xu, Dusit Niyato et al. "Toward edge general intelligence with agentic AI and agentification: Concepts, technologies, and future directions." arXiv preprint arXiv:2508.18725 (2025).

S. Ilager et al., Proteus: Towards Intent-driven Automated Resource Management for Edge Sensor Nodes, in 14th Workshop on AI and Scientific Computing, 2024.

Compétences

Cloud Computing, Edge Computing, Distributed systems

Python, Docker/Kubernetes

Reinforcement learning

Knowledge of sequential decision-making (MDP, influence diagrams, bandits, etc.)

Be independent and curious about research

Have an excellent level in English (spoken and written)

Multi-agent systems and/or Knowledge of the Grid '5000 testbed are a plus.

Avantages

- 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