Professor and lecturer at HE-Arc Engineering

Born in 1973

Lives in Neuchâtel

Nuria Pazos was born in Madrid. She studied at the Polytechnic in her home city and became an industrial engineer. In 1998, after a series of internships and a year of professional experience, she moved to Munich to study for her doctorate, where she worked in an integrated circuit laboratory. She then relocated to Neuchâtel with her partner, who was writing his thesis at the Institut de Microtechnique de Neuchâtel (IMT). She completed a post-doctorate qualification at EPFL, between the microelectronic systems laboratory and the processor architecture laboratory. In 2008, she joined HE-Arc Engineering. Once she had mastered French, she began teaching in the institution's IT department, specialising in industrial and embedded IT. What she particularly appreciates about the applied research activities at HE-Arc Engineering is the structure of skills groups set up by Philippe Grize. From her point of view, this organisation strengthens collaboration between the various groups, enabling them to respond effectively to the various needs of the industrial fabric – including through direct mandates or Innosuisse projects. For over a year, she has devoted part of her time to the European Bonseyes project, a platform that aims to strengthen Europe's place in the artificial intelligence race. Interview.

How did the Bonseyes project come about ?

The idea for this European project arose two years ago. It was set up in close collaboration with nViso, a company based in Vaud, through a meeting in Lisbon as part of a networking event organised by the European H2020 Research Programme for ICT (Information and Communication Technologies).  nViso invited us to join the consortium for an European H2020 project. This was the first European H2020 project for the HE-Arc engineering IT stream. It was an incredible opportunity to be part of a consortium of this size, made up of 14 partners, half from the private sector and half from research.The project was launched in December 2016. Three skills groups were involved: the embedded computing systems group, the imaging group and the iteration technologies group. But the project did not stop there. We have just submitted a second H2020 request for a €20 million project, together with 64 European partners.

How will the Bonseyes project strengthen access to artificial intelligence (AI) in Europe ?

At the moment, the major players in artificial intelligence are Google, Facebook, Netflix etc. They have the capacity to collect information everywhere (Big Data) and thus to feed algorithms in order to generate very powerful AI models.  No small European company working alone will ever have enough data to face the GAFAs. With Bonseyes, we offer a marketplace in which different companies and research institutes can pool their data, models, learning methods, implementations and knowledge.

Bonseyes is also trying to find a solution to another challenge: the processing of data collected through the Internet of Things (IoT) ?

Yes. In an ideal scenario, data collected from the internet, for example, is processed with powerful machines. However, many IoT objects, like self-driving cars, are based on edge devices – directly integrated into the car in this case. However, the very powerful algorithms currently used in Big Data cannot run on these devices that have limited energy, computing capacity and memory. As a result, the Bonseyes project seeks to develop models that can be optimally deployed on these autonomous devices, with real-time constraints.

Why did you structure the project from the perspective of embedded systems (edge computing) ?

An embedded system is located at the boundary between the data network and the device. It enables us to process data about an object that contains no intrinsic data at all, such as a microphone for example. By combining it with an intelligent system embedded directly on the object, the information can be processed on the spot, without being sent elsewhere, to a central server for example. The confidentiality of the data is thus guaranteed, especially when it comes to images or videos.

How will you meet the challenge of limited energy resources on embedded systems ?

To stay connected and avoid problems related to the heating of components, an embedded system must consume as little energy as possible. The energy consumption of materials can be limited, using energy-efficient components. But it can also be limited at software level: the simpler an algorithm, the less energy it consumes. It's up to us, embedded computing engineers, to make it as efficient as possible, without losing analysis power and accuracy. 

Written by Victoria Barras

Other articles


Associate professor of microtechnology at the EPFL
Holder of the Patek Philippe Chair and director of the micromechanical and horological design laboratory (INSTANT-LAB) in Neuchâtel.

Born in 1973

Lives in Neuchâtel

Read more


Incubated by Neode

Founded in 2017

8 employees

Read more


Tenure Track Assistant Professor at the EPFL Neuchâtel Branch

Born in 1974

Lives in Bern

Read more