Data Scientist, Deciphex

• A real research track record — a PhD in a quantitative or biomedical field, or equivalent hands-on research experience in industry
• Machine learning applied to biomedical data, ideally biomedical imaging
• Confidence with scientific literature: you can read into a new area and turn what you find into something you can test
• The full research workflow — scientific question, dataset, analysis, validation, and presenting the result
• Solid applied statistics and reproducible analysis practice
• Python and/or R, plus SQL
• Clear scientific communication: you can explain methods, limitations and conclusions to pathologists, engineers and external partners alike

Desirable

• There are very few places outside academia where you get to run your own research. This is one of them
• Deciphex builds AI for digital pathology. We're hiring a Data Scientist into our AI R&D team to take ownership of research that is currently waiting for someone to own it — including building the preclinical pathology benchmark datasets that don't yet exist anywhere publicly, and the scientific work behind virtual control groups for animal studies
• You'll have access to millions of whole-slide images and the resources to do something serious with them, working alongside pathologists and AI researchers who know the domain deeply
• Some of this work sits within two major EU-funded research partnerships, BIGPICTURE and VICT3R, so your results reach well beyond Deciphex — and contribute directly to the 3Rs in animal research
• Building benchmark datasets and evaluation tasks for pathology foundation models, in a space where no public benchmarks currently exist
• Statistical characterisation of control datasets — trend analysis, confounders, and batch effects across sites and scanners
• Curating and quality-controlling large whole-slide image datasets and their associated study metadata
• Validating downstream pathology AI models and building the evidence to stand behind them
• Working directly with pathologists to ground your datasets and results in real tissue morphology
• Publishing and presenting your work