Location Munich, Bavaria, Germany
Job ID R-226795
Date posted 14/05/2025
AstraZeneca is a global, science-led, patient-focused biopharmaceutical company that focuses on the discovery, development and commercialisation of prescription medicines for some of the world’s most serious diseases. But we’re more than one of the world’s leading pharmaceutical companies.
SITE DESCRIPTION - MUNICH, GERMANY
Welcome to Computational Pathology Munich, one of over 400 sites here at AstraZeneca, providing a collaborative environment where everyone feels comfortable and able to be themselves is at the core of AstraZeneca’s priorities, it’s important to us that you bring your full self to work every day. To help you maintain your best self, here’s a sneak peek into some of the things this site provides for you: After-work events, lunch & learns, spacious environment, sustainable office working environment, events, family and childcare support and of course the Alps around the corner for hiking, biking and skiing.
JOB DESCRIPTION
- Serving the goals of drug development projects, create image analysis solutions to generate decisive data based on the analysis of H&E, immunohistochemistry, and multiplexed immunofluorescence histopathology whole slide images
- Leadership role in the research, development and application of latest artificial intelligence methods
- Assume an image analysis leadership role in projects or segments of large-scale drug development programs
- Disseminate knowledge, best practices and developed algorithms and software tools to internal audiences with project meetings, seminars and documentation as well as to the external academic community with conference presentations and journal publications
- Collaborate closely with project stakeholders and interfacing functions such as translational science, pathology, quality assurance, data management, software development, data science and bioinformatics
- Comply with the AstraZeneca General Laboratory Standard to support product quality and regulatory compliance in the development, manufacture and distribution of Active Pharmaceutical Ingredients, Medicinal Products and Devices
QUALIFICATIONS:
- Ph.D. degree in computer sciences, mathematics, physics, bioinformatics or comparable degree with a focus on deep / machine learning and computer vision; or equivalent experience e.g. 3-5 years in industry
- Additional working experience in the field of medical and/or biomedical image analysis as post-doc or in industry
- Expert knowledge of state-of-the-art deep learning models and architectures for computer vision in particular for object detection, semantic segmentation, instance segmentation and weakly-supervised learning
- Advanced programming skills in Python and good knowledge of the most common scientific computing libraries (e.g. numpy, pandas, scikit-learn, scipy, opencv) and machine learning frameworks (e.g. tensorflow, pytorch).
- Good knowledge of best practices for software development (e.g. docstring, unit testing, CI pipelines) as well as for the development, validation and release of models (e.g. data splitting, model testing, hyperparameter tuning)
- Proven track record of scientific publications, conference contributions (e.g. NeurIPS, CVPR, ICML) and/or delivered AI solutions in scientific industry projects
- Cability to assume accountability for sub-project management and delivery in a multi-national, multi-disciplinary project in academia or industry
- Talent to conceptualize, build, communicate and implement novel scientific ideas aligned with the drug development and biomarker strategies
- Experience with training and deploying computer vision foundation, whole-slide image representation learning and/or multimodal fusion of different (imaging) modalities using state-of-the-art approaches (e.g. CLIP) is a strong plus
- Working experience with at least one of the following: semi-supervised learning, self-supervised learning, few-shot learning, generative learning, self-attention learning (e.g. Transformers), active learning, survival learning, graph neural networks and / or domain adaptation learning.
Linked-in
Google