Software Engineer (M/F) – Internship

Why this role is key

We want to explore the integration of neural networks directly into our antenna systems. Our aim is to integrate our models so that they run efficiently in our embedded systems. Working at the interface between the Software and Algorithm & Modelling teams, you will contribute to deployment and compare different hardware acceleration solutions, assessing their performance, power consumption and model accuracy.

Reporting to Louis, Software Engineer, this internship will allow you to bridge artificial intelligence and embedded systems, from creating a common software interface for multiple accelerators to running models on a target platform. You will also learn to adapt a neural network to hardware constraints, in collaboration with the Algorithm & Modelling team.

Your responsibilities

Getting started and development

• Familiarise yourself with two hardware platforms and run an initial neural network on each,

• Develop a common C++ interface to run models on different AI accelerators.

Platform testing and comparison

• Set up tests to compare processing speed, energy consumption and model accuracy across the different platforms,

• Analyse the results to identify the capabilities and limitations of each solution.

Antenna model deployment

• Integrate the models and run them on the embedded platform,

• Adapt the models to hardware constraints with support from this team, then measure the impact on their accuracy,

• Deliver a demonstration on the target platform.

Documentation and reporting

• Document the development work and provide usage examples,

• Present the comparison results and the work completed.