The Position
Are you interested in speech signal processing, machine learning, and analyzing digital health datasets for signs of disease progression? Do you want to apply your knowledge to real-world data, and impact the lives of people living with neurological diseases? Then come help us apply deep learning methods to develop better digital measures of speech impairment!
Roche is seeking candidates either pursuing a PhD degree or recent graduates from a Master degree, with expertise in signal processing, machine learning (deep learning) and explainable artificial intelligence (XAI) to develop new features that help track disease progression.
In this internship, you will apply the latest deep learning methods to rich, real-world speech datasets collected by the Digital Biomarkers team in the scope of Roche's clinical studies from patients with neurological diseases, and assess what drives the network predictions using XAI methods, helping the Data Analysis team identify novel acoustic markers of disease progression.
Responsibilities:
Design and build neural embedding extraction pipelines using open-source toolkits and network implementations.
Build and compare regression models of disease progression using both existing, traditional acoustic features developed by the Digital Biomarkers Data Analysis team, as well as neural embeddings.
Using XAI techniques, investigate the underlying speech properties that are captured by the neural embeddings, and work with team members to design new acoustic features based on these insights.
In collaboration with the data analysis team, design experiments, evaluate performance, discuss results and refine features, with the goal of ultimately developing improved, speech-based measures of disease progression.
Experience and Competencies Preferred:
You possess an in-depth understanding of neural networks theory, signal processing, and feature selection. Knowledge of speech processing techniques is a big plus.
You are proficient in Python and experienced with related scientific computing packages (knowledge of NumPy, scikit-learn, Matplotlib and PyTorch is a must). Experience with distributed/grid and/or cloud computing environments is a plus.
You have experience with classical machine learning techniques (supervised learning, training, validation, performance metrics).
You have hands-on experience working with deep neural network frameworks and you feel comfortable experimenting with new approaches using open-source tools and resources. Experience in eXplainable AI is a big plus.
You possess good analytical skills and attention to detail. You critically question your results, and are willing to go the extra mile to verify key assumptions
Last but not least, you enjoy all of the above! You have a real passion to apply new techniques to real-world data, like discussing your insights with your colleagues, and feel motivated to uncover new features of speech that will impact the lives of patients.
Qualifications Required:
MSc graduate or PhD degree candidate in Machine Learning/Signal Processing related field (e.g., Computer Science, Bioinformatics, Electrical or Electronics Engineering).
Proficient in English
Preferred Start Date: May (1-2 months flexible)
Workload: 100%
Duration: 6 months
Have we piqued your interest? Then we look forward to receiving your complete application documents (letter of motivation, CV, job references, diplomas, etc.)
Due to regulations non-EU/EFTA citizens must provide a certificate from the university stating that an internship is mandatory as part of the application.
Who we are
At Roche, more than 100,000 people across 100 countries are pushing back the frontiers of healthcare. Working together, we've become one of the world's leading research-focused healthcare groups. Our success is built on innovation, curiosity and diversity.
Basel is the headquarters of the Roche Group and one of its most important centres of pharmaceutical research. Over 10,700 employees from over 100 countries come together at our Basel/Kaiseraugst site, which is one of Roche`s largest sites. Read more.
Besides extensive development and training opportunities, we offer flexible working options, 18 weeks of maternity leave and 10 weeks of gender independent partnership leave. Our employees also benefit from multiple services on site such as child-care facilities, medical services, restaurants and cafeterias, as well as various employee events.
We believe in the power of diversity and inclusion, and strive to identify and create opportunities that enable all people to bring their unique selves to Roche.
Roche is an Equal Opportunity Employer.
Job Level:
Entry Level
