ML Research Engineer / Scientist (Remote, international)

Role overview This is a research engineering and applied science role focused on building foundation models that interpret full radiological studies rather than narrow, single-finding detections. The work sits on a short path from training to patient impact, with models advancing through regulatory clearance and into live clinical reading rooms. It suits a self-directed researcher who wants end-to-end ownership of ideas, experiments, and the systems that ship.

Responsibilities - Design and train foundation models that process a complete volumetric CT study in a single pass rather than slice by slice - Build vision-language components that connect image features to the structured language radiologists write in reports - Develop a single model capable of flagging multiple urgent conditions at once, calibrated for safe use alongside a radiologist - Run large-scale training, evaluation, and ablation experiments on production-scale medical imaging data - Drive projects from initial hypothesis through to the operating point that ships in regulatory submissions and hospital deployments - Collaborate with engineering and clinical specialists to ensure models meet reproducibility, safety, and clinical-readiness standards

Requirements - Strong proficiency with PyTorch, including writing custom architectures, training loops, and distributed training code rather than only calling high-level APIs - Solid grasp of the mathematical reasoning behind loss functions and architectures, not just their usage - Track record of designing and running independent experiments with minimal direction - Commitment to rigorous evaluation, reproducibility, and documentation because model behavior carries clinical weight - Comfort working remotely with a distributed, cross-functional team

Nice to have - Background in medical imaging, 3D data, or other volumetric modalities - Experience with vision-language models or self-supervised learning - Familiarity with DICOM, CT, or radiology workflows - A PhD or peer-reviewed publications, treated as a plus rather than a hard requirement

Benefits and work setup - Fully remote, open to candidates worldwide - Cash-weighted base salary calibrated to the local market of the country where the work is performed, with the specific range shared early in the process - International offers are cash only with no equity component, communicated transparently up front - Small, shipping-oriented team where research output reaches regulatory submissions and patient deployments within months - Opportunity to learn clinical radiology directly from practicing, fellowship-trained clinicians embedded with the modeling team