AI/ML models for flow scheduling problem

Nokia; Massy; Sur place; EUR 24 000 - 36 000; Plein temps

Résumé du poste

Join Nokia's innovative team to work on AI/ML models addressing the flow scheduling problem, utilizing advanced techniques like graph neural networks and LLMs. You will have the opportunity to contribute to groundbreaking research while gaining valuable experience in a collaborative, inclusive environment.

Prestations

Flexible working schemes

Continuous learning opportunities

Well-being programs

Mentoring programs

Inclusive culture

Qualifications

• Solid experience in ML and neural networks required.
• Good programming skills in Python needed.
• Knowledge of operations research is advantageous.

Responsabilités

• Develop and compare AI/ML models for flow scheduling problems.
• Implement advanced ML techniques in industrial use cases.
• Learn about deterministic networks and optimize scheduling.

Connaissances

Machine Learning

Python programming

Neural networks

Optimization techniques

Outils

PyTorch

Tensorflow

Keras

Description du poste

Social network you want to login/join with:

AI/ML models for flow scheduling problem, Massy

col-narrow-left

Client:

Location:

Massy, France

Job Category:

Other

-

EU work permit required:

Yes

col-narrow-right

Job Reference:

6dbdc9f13595

Job Views:

5

Posted:

12.07.2025

Expiry Date:

26.08.2025

col-wide

Job Description:

Join us in creating the technology that helps the world act together

We are a B2B technology innovation leader pioneering the future where networks meet cloud. At Nokia you will have a positive impact on people’s lives and help build the capabilities needed for a more productive, sustainable, and accessible world.

Be part of a culture built on an inclusive way of working where we are open to your ideas, you are empowered to take risks and are encouraged to be fearless in bringing your authentic self to work.

The team you'll be part of

Nokia Bell Labs is the world-renowned research arm of Nokia, having invented many of the foundational technologies that underpin information and communications networks and all digital devices and systems. This research has produced nine Nobel Prizes, five Turing Awards and numerous other awards.

Our research department involves 12 highly skilled researchers in France in US. In multi-cultural environment your work will contribute to solve the delay problem. We have proposed the concept of Strict Deterministic Networks (St-DetNet), a packet network technology that delivers data within strict deadlines over IP/Ethernet networks. A St-DetNet schedules data frames in periodic deterministic streams, avoiding buffering processes as much as possible. And in cases where buffering cannot be avoided, it ensures that the corresponding delays are known exactly in advance. In short, the position of each data frame in the network is known exactly at each clock cycle.

On a network made up of several nodes, the design of a St-DetNet solution requires the complete specification of the time steps at which each data frame passes through each node. This flow scheduling problem is unfortunately NP-hard and can therefore only be solved perfectly for small instances. However, based on a limited number of experiments, we believe that AI/ML models may be able to infer relevant solutions.

The objective of this internship is to develop, experiment with and compare several AI/ML models for this flow scheduling problem.

ML methods will then learn to construct scheduled data flows for different network topologies and strict time constraints. Several approaches are available, for example: graph neural networks (GNNs) and large language models (LLMs). In the first case, which is the most natural, a GNN would be designed to learn programmed data streams based on graphical descriptions of input networks and time constraints. In the second case, the scheduling problem would be reformulated textually - with the task description given as a prompt and the problem instance encoded as a string - and then passed to an LLM with the aim of learning the admissible scheduled data streams.

What you will learn and contribute

Are you passionate about problem solving?

As a member of our team, you will learn new concepts such as deterministic networks,

implement the most advanced Machine Learning techniques on industrial use cases

Your skills and experience

Solid experience in ML and neural networks

Good programming skills, preferably in Python: experience with neural network libraries such as PyTorch, Tensorflow or Keras is highly recommended.

It would be great if you also had knowledge of operations research, optimisation techniques and previous experience of scheduling problems would be a real advantage.

What we offer

Nokia offers flexible and hybrid working schemes, continuous learning opportunities, well-being programs to support you mentally and physically, opportunities to join and get supported by employee resource groups, mentoring programs and highly diverse teams with an inclusive culture where people thrive and are empowered.

Nokia is committed to inclusion and is an equal opportunity employer

Nokia has received the following recognitions for its commitment to inclusion & equality:

• One of the World’s Most Ethical Companies by Ethisphere
• Gender-Equality Index by Bloomberg
• LGBT+ equality & best place to work by HRC Foundation