Director, Predictive and Agentic Target Biology

Responsibilities

• Take pre-trained biological foundation models and adapt them to target biology questions with scientific rigour, fine-tuning, probing, and applying models (e.g. transcriptomic, high-content imaging) to questions such as target-disease mechanism.
• Apply and evaluate virtual-cell and in-silico perturbation approaches to predict cellular response and support mechanistic, in-silico target discovery.
• Integrate heterogeneous target evidence (omics, functional genomics, causal biology, chemogenomics, literature, clinical) into structured, traceable representations, applying knowledge-graph and graph-learning methods.
• Design and build agentic scientific workflows that plan, retrieve, analyse, and synthesise across the target discovery process, using current orchestration patterns (multi-agent systems, RAG, MCP-based tool access) with grounding, provenance, and rigorous evaluation of agent reliability.
• Evaluate emerging AI/ML approaches for discovery data science, benchmark their performance with appropriate validation strategies, and implement them into reliable, production use.
• Apply biological and chemical domain knowledge to ensure models are scientifically grounded, interpretable, and address meaningful therapeutic questions.
• Partner with scientific domain experts to ensure predictive and agentic methods address the questions that matter most to the portfolio.
• Contribute to the MLOps practices that bring predictive methods into reliable use, including model deployment, monitoring, and drift detection for robust model serving.

Basic Qualifications

• Master's degree (or higher) in Computer Science, Computational Biology, Bioinformatics, or a related field.
• 15+ years of software engineering and applied computational science experience in bioinformatics, cheminformatics, or scientific computing.
• Experience in pharmaceutical, biomedical, or life sciences preclinical discovery data.
• Experience applying machine learning, graph-based, or AI agent methods to biological or chemical data.
• Experience programming in Python and/or R, including machine learning libraries such as scikit-learn, PyTorch, or scverse.
• Experience with production AI/ML systems and MLOps practices (CI/CD, containerisation, cloud deployment).

Preferred Qualifications

• Fine-tuning or applying biological foundation models to target biology questions.
• Virtual-cell, in-silico perturbation, or perturbation-prediction methods, with an understanding of how to validate their reliability.
• Building or deploying agentic workflows (multi-agent orchestration, RAG, MCP-based tool access, foundation-model integration).
• An instinct for matching the right method to the scientific question, and for what makes a prediction credible enough to act on.
• Knowledge-graph or graph-learning methods for integrating heterogeneous evidence.
• Productionising ML models, model monitoring, and automated retraining; hands-on MLOps (versioning, experiment tracking, containerisation).
• Deep-learning architectures for biological data (transformers, diffusion models, GNNs), and data engineering concepts (ETL, databases, data lakes).
• Strong communication skills, with the ability to translate between biological and technical audiences.