Associate Principal AI Research Scientist (Fundamental AI Research for Digital Biology)

Overview This role is based in Barcelona, with an on-site commitment of three days a week. Fluency in English is required.

Introduction to the opportunity Are you passionate about creating artificial intelligence and machine learning systems for real-world science applications? Does contributing to preventing, modifying, and even curing some of the world's most complex diseases inspire you? Would you like to work on developing an iterative drug discovery and development process while drawing on methods across various fields, from active learning to optimisation and search? What about advancing our understanding of biology, streamlining research and development processes, and leveraging a variety of data modalities? Do you thrive working in a supportive, inclusive environment where creativity, collaboration across disciplines and lifelong learning towards innovative breakthroughs are encouraged? If yes, this opportunity may be for you.

Join our interdisciplinary Centre for Artificial Intelligence team working on the frontier of AI research for digital biology. Your work will support the next generation of medicines and vaccines at the intersection of AI, biology, and engineering. Your work will contribute to transforming the drug discovery and development value chain as we know it, uncovering novel biological insights, automating processes, streamlining decisions, and improving the overall pipeline across all therapeutic areas at AstraZeneca.

Accountabilities You will work efficiently in a team to lead and deliver projects optimally, researching, developing and using the novel AI theories, methodologies, and algorithms, with engineering best practices and standard processes for various biology, chemistry and clinical applications.

You will be part and also lead multifunctional projects to conceive, design, develop and conduct experiments to test hypotheses, validate new approaches, and compare the effectiveness of different AI / ML systems, algorithms, methods and tools for new applications to support the discovery, design, and optimisation of medicines with improved biological activity.

You will lead and contribute to addressing challenges and opportunities in the drug discovery and development value chain processes and provide innovative solutions in fields such as deep learning, representation learning, reinforcement learning, meta-learning, active learning approaches applied to de novo molecule design, protein engineering, in-silico discovery, structural biology, computational biology, translational sciences, biomarker discovery, clinical research, clinical trials and many other areas.

You will lead and develop machine learning models designed explicitly for analysing heterogeneous biological data while collaborating with biology researchers to run algorithmically designed wet lab experiments to inform future experimental directions.

You will remain at the forefront of AI / ML research by participating in journal clubs, seminars, mentoring, and personal development initiatives and contributing to publications and academic and industry collaborations.

Essential Skills / Experience A PhD in machine learning, statistics, computer science, mathematics, physics, or a related technical discipline with relevant fundamental research experience in artificial intelligence and machine learning or equivalent practical experience.

Fundamental AI research experience in conjunction with foundational knowledge and a proven track record in conceptualising, designing, and creating entirely new models, methods, approaches, architectures, and algorithms from scratch. This is essential as off-the-shelf methods and state-of-the-art AI / ML techniques often do not work on our scientific problems and datasets.

Deep theoretical understanding, combined with a strong quantitative knowledge of algebra, algorithms, probability, calculus, and statistics, as well as extensive hands-on experimentation analysis, and AI / ML techniques visualisation.

Well-rounded experience designing new AI / ML approaches to deriving insights from proprietary and external datasets to generate testable hypotheses using algorithmic, mathematical, computational, and statistical methods combined with theoretical, empirical or experimental research sciences approaches.

Experience in theoretical, fundamental AI research and practical aspects of AI / ML foundations and model design, such as improving model efficiency, quantisation, conditional computation, reducing bias, or achieving explainability in complex models.

In-depth understanding of applying rigorous scientific methodology to ( i ) identify and create novel ML techniques and the required data to train models , (ii) develop machine learning model' architectures and training algorithms, (iii) analyse and tune experimental results to inform future experimental directions, and (iv) implement and scale training and inference engineering frameworks and (v) validate hypotheses.

Distinctive experience in exploit ing the simplest tricks to the latest research methods to advance AI / ML capabilities while implementing them in an elegant, stable, and scalable way.

Thorough a lgorithmic development and programming experience in Python or other programming languages and standard machine learning toolkits, especially deep learning (e.g., Pytorch , TensorFlow, etc.).

Robust ability to communicate and collaborate effectively with diverse individuals and functions, reporting and presenting research findings and developments clearly and efficiently to other scientists, engineers and domain experts from different disciplines.

Fundamental research, extensive research and expert understanding combined with hands-on practical experience and theoretical knowledge of at least two or more of the following research areas - examples include but are not limited to - multi-agent systems, logic, causal inference, Bayesian optimisation, experimental design, deep learning, reinforcement learning, non-convex optimisation, Bayesian non-parametric, natural language processing, approximate inference, control theory, meta-learning, category theory, statistical mechanics, information theory, knowledge representation, unsupervised, supervised, semi-supervised learning, computational complexity, search and optimisation, artificial neural networks, multi-scale modelling,