Industrial PhD-AI Training and Deployment under Sec Constraints

Industrial PhD - AI Training and Deployment under Security Constraints



About

The PhD position is part of an industrial collaboration between Saab AB, Software Center at Chalmers University of Technology and University of Gothenburg, and WASP (Wallenberg AI, Autonomous Systems and Software Program).



Your role

Successful machine- and deep-learning models require large amounts of data from real-world deployments for training and validation purposes. In some cases, such as in the defense industry, collecting this data to a central location is unfeasible or undesirable because of security and confidentiality reasons, excessive communication cost, or other reasons.



In addition, a centrally trained model may not suit all customers as different customers may have significantly varying contexts and circumstances and, consequently, want their ML/DL model deployment to be optimized for their specific situation with a reduced amount of data.



Conceptually, we would want a situation where a model can be trained up to some extent

centrally using the available data, then the model is deployed to all customers and locally trained based on customer data and then the learnings of all the deployed models are made available again centrally. This would allow us to build a solution where all customers are enjoying continuously improving systems by sharing their learnings with each other and receiving the combined learnings in updated ML/DL models. And all of this while maintaining confidentiality, limiting communication cost and allowing for customization of each system

deployment for specific customer needs.



From a more theoretical perspective, ​Transfer Learning (TL) ​provides part of the solution, but it only allows deployment of a centralized model to local contexts and training. To create

industrial strength, production-quality solutions, research is needed to address three main challenges:



● We need solutions that support the merging of local training results with the central model without sacrificing confidentiality and security.



● We need solutions that allow for the deployment of updated centrally trained models to customers who maintain the local customization that was trained into the previous deployment of the model.



● We need to investigate the required engineering infrastructure, necessary ways of working (both internally and with customers) and deployment approaches that maintain the security requirements of defense customers.



Your profile

* Master's level degree corresponding to at least 240 higher education credits
* Very good verbal and written communication skills in Swedish and English
* Interest and skills in AI
* Interest and skills in SW development



This position requires that you will be approved in a security screening in accordance with the Swedish Protective Security Act.



Note: In order to start this Industrial cooperation between SAAB AB and Software Center at Chalmers University of Technology and University of Gothenburg, we have to apply for

funding for a named applicant. Regardless the outcome of this application, you will be

offered a position at Saab.



What you will be a part of

Surveillance, a Business Area within Saab, is a world-leading supplier of systems for detection of threats and self-protection. Business Unit Radar Solutions is responsible for Radar in airborne-, surface- and naval systems. This position will be organized within our Innovation Lab.



If you aspire to help create and innovate whilst developing yourself in a challenging team

setting, Saab may well have the perfect conditions for you to grow. We pride ourselves on a nurturing environment, where everyone is different yet we share the same goal - to help

protect people.



For questions, please contact:

Ulf Näsström, Recruiting Manager

+46 734 379 051



Sven Nilsson, Technical Specialist

+46 734 379 410


Åsa Mårtensson, HR Specialist Recruitment

+46 10 2165107