Post-Doctoral Research Visit F/M Model placement in inference delivery networks

Inria, the French national research institute for the digital sciences; France; Sur place; EUR 35 000 - 50 000; Plein temps

Résumé du poste

A leading research institute in France is seeking a full-time postdoc to explore AI model placement within intelligent networks. Responsibilities include conducting research on optimization techniques and writing academic papers. Candidates should have a PhD in Computer Science or a related field and fluency in English is essential. Offers include flexible working hours and substantial leave.

Prestations

Partial reimbursement of public transport costs

7 weeks of annual leave plus 10 extra days off

Possibility of teleworking

Professional equipment available

Social, cultural, and sports events

Qualifications

• Basic level in French is required.
• Good level in English is essential.

Responsabilités

• Study AI model placement in IDNs.
• Optimize trade-offs between model effectiveness and resource availability.
• Evaluate methodologies for inference quality.

Connaissances

Read and synthesize literature work

Conducting cutting-edge research

Proposing novel approaches

Writing research papers

Disseminating research findings

Understanding of networking principles

Programming languages proficiency

Algorithm design and implementation

Familiarity with optimization techniques

Excellent communication skills

Team collaboration

Formation

PhD in Computer Science or related field

Description du poste

Inria, the French national research institute for the digital sciences

Organisation/Company Inria, the French national research institute for the digital sciences Research Field Computer science Researcher Profile Recognised Researcher (R2) Country France Application Deadline 31 Dec 2025 - 00:00 (UTC) Type of Contract Temporary Job Status Full-time Hours Per Week 38.5 Offer Starting Date 1 Jan 2026 Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Reference Number 2025-08759 Is the Job related to staff position within a Research Infrastructure? No

Offer Description

This PostDos is funded by the c hallenge Inria-Nokia Bell Labs: LearnNet (Learning Networks)

Assignments :

In this postdoc, we will study the problem of AI model placement in an IDN. This is a challenging optimization problem that involves a non-trivial tradeoff between model effectiveness, inference latency, and resource availability while also dealing with the natural dynamicity of the network, e.g., due to users’ request process or changes in available computing and communication resources.

We will also consider other metrics, such as energy consumption, in the objective functions and networking constraints for systems where the network presents some inelasticity (see also [1]). We will leverage multi-objective optimization techniques (e.g., Pareto efficient solutions) and transfer learning techniques to adapt models across nodes with different levels of knowledge and resource availability. We will also rely on online learning approaches to achieve model placements with adversarial guarantees regarding regret.

In comparison to our preliminary work in[2] or [3], we will allow models to be split across multiple nodes [4,5,6]. In particular, we aim to compare specific model splitting techniques, with or without the insertion of bottlenecks[7,8] (reference[8] is also the result of NEO-AIRL cooperation), in terms of performance metrics like inference delay and network load. We will evaluate different methodologies to estimate online the quality of an inference [9].

This evaluation may also consider scenarios with significant heterogeneity of the nodes, such as in the scenario of embedded Edge AI or even more with TinyML (resources possibly lower by orders of magnitude but possibly a massive number of devices).

This postdoc will be recruited and hosted at InriaSaclay and supervised by Tribe (INRIA), Neo (INRIA), and AIRL (Nokia)

References:

[1] Kinda Khawam et al. “Edge Learning as a Hedonic Game in LoRaWAN”. ICC 2023 - IEEE International Conference on Communications.2023.

[2]Tareq Si Salem et al. “Towards inference delivery networks: distributing machinelearning with optimality guarantees.” In: 19th Mediterranean Communicationand Computer Networking Conference (MEDCOMNET 2021). Ibiza(virtual), Spain: IEEE, June 2021, pp. 1–8.

[3]Wassim Seifeddine, Cédric Adjih, and Nadjib Achir. “Dynamic HierarchicalNeural Network Offloading in IoT Edge Networks.” In: PEMWN 2021 - 10thIFIP International Conference on Performance Evaluation and Modeling inWireless and Wired Networks. Ottawa / Virtual, Canada: IEEE, Nov. 2021,pp. 1–6.

[4]Surat Teerapittayanon, Bradley McDanel, and Hsiang-Tsung Kung. “Branchynet:Fast inference via early exiting from deep neural networks”. In: 2016 23rd InternationalConference on Pattern Recognition (ICPR). IEEE, 2016, pp. 2464-2469

[5] S. Teerapittayanon, B. McDanel, and H. T. Kung. “Distributed Deep NeuralNetworks Over the Cloud, the Edge and End Devices”. In: 2017 IEEE 37thInternational Conference on Distributed Computing Systems (ICDCS). ISSN:1063-6927. June 2017, pp. 328–339.

[6] Yoshitomo Matsubara, Marco Levorato, and Francesco Restuccia. “Split Computingand Early Exiting for Deep Learning Applications: Survey and ResearchChallenges”. In: ACM Computing Surveys 55.5 (Dec. 2022), 90:1–90:30. issn:0360-0300.

[7] Yoshitomo Matsubara et al. “BottleFit: Learning Compressed Representationsin Deep Neural Networks for Effective and Efficient Split Computing”. English.In: IEEE Computer Society, June 2022, pp. 337–346. isbn: 978-1-66540-876-9.

[8] Gabriele Castellano et al. “Regularized Bottleneck with Early Labeling”. In: ITC2022 - 34th International Teletraffic Congress. Shenzhen, China, Sept. 2022.

[9] Ira Cohen and Moises Goldszmidt. “Properties and benefits of calibrated classifiers”.In: European Conference on Principles of Data Mining and KnowledgeDiscovery. Springer, 2004, pp. 125–136.

• Read and synthesize literature work,
• Conducting cutting-edge research at the intersection of networking and AI
• Propose novel approaches and technical solutions for AI model placement
• Writing research papers for submission to top-tier conferences and journals in networking, AI, and computer science.
• Disseminating research findings through presentations at conferences, seminars, and workshops.
• A solid understanding of networking principles, protocols, and architectures is essential.
• Proficiency in programming languages commonly used in AI and networking research.
• Experience with relevant libraries and frameworks is also valuable.
• Ability to design and implement algorithms for solving complex problems.
• Familiarity with optimization techniques.
• Excellent written and verbal communication skills for presenting research findings, writing academic papers, and collaborating with peers.
• The ability to work effectively as part of a research team, collaborate with colleagues from diverse backgrounds, and contribute positively to group dynamics

Languages FRENCH Level Basic

Languages ENGLISH Level Good

Additional Information

• 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 and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities