Senior Machine Learning Engineer / Research Scientist – Healthcare LLMs & Clinical Risk Prediction

About the Role We are building a next-generation AI model for the healthcare industry, inspired by models such as Delphi-2M, to predict long-term disease risks from longitudinal electronic health records (EHRs). This role focuses on designing, training, and validating transformer-based models on structured and temporal healthcare data (not just text). The ideal candidate has experience in sequence modeling, healthcare ML, and large-scale model implementation. Key Responsibilities Design and implement transformer-based models for longitudinal healthcare data (EHRs, ICD codes, lab events, demographics). Convert clinical records into tokenized temporal sequences (similar to language modeling). Train large-scale models for multi-disease risk prediction and time-to-event forecasting. Develop probabilistic outputs (risk scores, survival curves, disease timelines). Evaluate models using healthcare-specific metrics (AUC, calibration, time-dependent C-index). Address bias, fairness, and explainability in clinical predictions. Collaborate with clinicians, data scientists, and product teams to align model outputs with real-world healthcare needs. Document methodology for regulatory, ethical, and clinical review. Required Skills & Experience Machine Learning & Modeling Strong experience with deep learning and transformers (PyTorch / JAX / TensorFlow). Experience with sequence modeling (time-series, event sequences, survival analysis). Familiarity with architectures inspired by LLMs applied beyond text. Experience training models on large datasets (100k+ records). Healthcare / Clinical Data Hands-on experience with EHR data, claims data, or clinical registries. Knowledge of ICD-10 / SNOMED / CPT codes and medical ontologies. Understanding of longitudinal patient data and disease progression modeling. Data Engineering Data preprocessing pipelines for temporal, sparse, and noisy healthcare data. Feature/token engineering for clinical events. Experience with distributed training, GPUs, and cloud platforms. Nice to Have (Strong Plus) Experience with Delphi-2M, BEHRT, Med-BERT, RETAIN, or similar models. Background in biostatistics, epidemiology, or survival analysis. Publications in healthcare ML, AI in medicine, or applied ML conferences. Familiarity with UK Biobank, MIMIC, or national health registry datasets. Knowledge of healthcare compliance (HIPAA, GDPR, anonymization). What We’re Building A multi-disease risk prediction model covering hundreds to thousands of conditions. A foundation model for preventive and predictive healthcare. AI that supports clinicians — not replaces them — with interpretable insights. Who Should Apply ML Engineers wanting to move beyond generic LLMs into real-world impact. Research Scientists interested in transformers for healthcare & life sciences. Candidates excited by long-term disease modeling, ethics, and clinical AI. Apply tot his job