استخدام کارشناس علم داده (حوزه پزشکی-دورکاری) در ایزورا

What you’ll do - Develop, train, and validate ML/AI models on healthcare and clinical datasets. - Work with real-world clinical data — EHR, imaging, labs, registries — including messy, incomplete, and multi-source data. - Design rigorous validation: clinical relevance, bias, generalization, calibration, and safety; document the evidence. - Translate clinical workflows and questions into well-scoped modeling problems with clinicians. - Build reproducible data and modeling pipelines with full traceability and auditability. - Partner with clinical advisors and engineering to move models toward governed, production use. - Uphold data privacy, security, and regulatory expectations (including PHI handling). What we’re looking for - Strong experience applying AI/ML to healthcare data and clinical problems. - Hands-on with medical datasets (clinical / EHR, imaging, labs, or registries) and their real-world pitfalls. - Solid understanding of clinical workflows and how models fit into care. - Rigorous model validation: study design, metrics, bias, generalization, and clinical safety. - Discipline around reproducibility, documentation, and data governance. Technical skill stack The tools and technologies you should be strong in. We don’t expect every single item — depth in the core stack matters most. - Programming: Python (expert) and SQL; R a plus - ML & deep learning: scikit-learn, PyTorch, TensorFlow / Keras, XGBoost / LightGBM - Data engineering: pandas, NumPy, Polars, SciPy, statsmodels; reproducible pipelines - Clinical NLP: Hugging Face Transformers, spaCy / scispaCy, cTAKES or MedCAT for clinical text - Medical imaging (role-dependent): MONAI, SimpleITK, pydicom, nibabel, OpenCV - Healthcare data standards: FHIR, HL7, DICOM, OMOP CDM; ICD, SNOMED CT, and LOINC coding - Statistics & validation: biostatistics, survival analysis, calibration, causal inference, bias / fairness metrics, study design - MLOps: MLflow, Weights & Biases, DVC, Docker; experiment tracking and model registries - Data platforms: PostgreSQL, BigQuery or Snowflake; Spark a plus - Cloud & AI platforms: AWS SageMaker, GCP Vertex AI, or Azure ML - Privacy & compliance: PHI / HIPAA handling, de-identification, and data governance (GDPR / PIPEDA aware)