Postdoctoral position in scientific computing - Hiring in process/Finished, not possible to apply

The Department of Information Technology provides education and research of the highest international quality. The department educates roughly 4,000 students each year, and houses about 30 research teams. The strong research focus impacts and provides an excellent foundation for undergraduate education in the department. The department is building on activities in IT that have been carried on at Uppsala University since the mid 1960s. More info: http://www.it.uu.se/.

At the Division of Scientific Computing, we conduct research in the entire chain of what is needed to perform simulations; to mathematically describe the phenomenon under investigation, to formulate a solution method to the mathematical problem, and finally to construct computer programs that efficiently implement the developed solution method to enable the simulation.

http://www.it.uu.se/research/scientific_computing

Research Project: Federated machine learning with application in radiation oncology.

You will be conducting your research as part of the Integrative Scalable Computing Laboratory (http://iscl.research.it.uu.se) led by Associate Professor Andreas Hellander and Assistant Professor Salman Toor. You will be expected to actively participate in the groups activities and contribute positively to the wider research environment. This position is part of the strategic research area effort eSSENCE´s PostDoc-program towards new e-science methods and tools for artificial intelligence in research.

The research will be conducted in close collaboration with the machine learning group at RaySearch Laboratories led by Dr. Fredrik Löfman. The group develops software to implement deep learning segmentation of medical image data and machine learning for radiotherapy treatment planning. The postdoc will be given the opportunity to spend time at RaySearch’s offices to interact with the team and to learn about the application domain.

This project has the dual goal of:

1. Unlocking sensitive training data to improve deep learning organ segmentation for radiation oncology.
2. Improving Federated Machine Learning (FedML) methodology. FedML is a new class of methods for training ML models with privacy-preservation.

There are many situations in which it is not possible to centralize data for training machine learning models, such as for regulatory reasons, because the datasets are too large, or because the data is sensitive. In those situations, federated machine learning allows participating parties to train a joint global model without moving or disclosing any local private data. The project aims at developing new methodology and a technology platform for federated machine learning. A key aim of this project is to push the boundaries for practical use of federated learning, its performance, scalability, and robustness, driven by applications in deep learning organ segmentation.

Radiation therapy(RT) planning software helps clinicians develop the complex treatment plans needed to accurately deliver radiation to cancer tumors while avoiding damaging tissue and organs. A critical part of the process is accurate segmentation of organs from 3D imaging of the patient (typically CT scans). Traditionally, this has been a manual and time-consuming step. Recently, automatic organ segmentation based on deep learning has shown great promise to improve both accuracy and segmentation speed. However, deep learning segmentation relies critically on expert-annotated clinical data from different body sites to train the models. The project aims to demonstrate how federated learning can unlock private, sensitive training data.

Duties: The position is focused on research in the above described research project but may include a limited amount of teaching and departmental duties, but not more than 20%.

Requirements: To qualify for an employment as a postdoctor you must have a PhD degree or a foreign degree equivalent to a PhD degree in Scientific computing, Computer science, Applied mathematics, Statistics, or a for the project relevant area. The PhD degree must have been obtained no more than three years prior to the application deadline. The three year period can be extended due to circumstances such as sick leave, parental leave, duties in labour unions, etc.

The successful candidate must have documented experience in advanced and applied machine learning and programming. Personal qualities such as dedication, motivation, initiative and independence are valuable. Fluency in spoken and written English is required.

Additional qualifications: Documented experience of:

• Distributed- and cloud computing
• Cyber security
• Image analysis
• Software engineering and large-scale software development.

How to apply: The application should include a cover letter (no longer than 2 pages) including a biography including a summary of the doctoral work in relation to state-of-the art. The application should also contain a list of credentials (CV), copies of relevant certificates and grades, a list of publications and contact information to at least three reference persons.

Uppsala University strives to be an inclusive workplace that promotes equal opportunities and attracts qualified candidates who can contribute to the University’s excellence and diversity. We welcome applications from all sections of the community and from people of all backgrounds.

Salary: Individual salary.

Starting date: As soon as possible but no later than 2020-09-01.

Type of employment: Temporary position for 2 years according to central collective agreement.

Scope of employment: 100 %

For further information about the position please contact: Associate Professor Andreas Hellander, Andreas.Hellander@it.uu.se.

Please submit your application by April 3, 2020, UFV-PA 2020/619.

Are you considering moving to Sweden to work at Uppsala University? If so, you will find a lot of information about working and living in Sweden at www.uu.se/joinus. You are also welcome to contact International Faculty and Staff Services at ifss@uadm.uu.se.