Mass General Brigham is seeking a Postdoctoral Research Fellow to contribute to cutting-edge research with real-world impact at the intersection of neuroscience, critical care, and computational modeling, focusing on developing advanced AI/ML models and frameworks for ICU medicine and ALS research.
Requirements
- Strong expertise in machine learning and deep learning applied to clinical/biomedical data.
- Proven experience with transformers, LLMs, and modern NLP frameworks.
- Proficiency in Python, PyTorch, and related ML toolchains; experience with EHR/ICU note preprocessing and feature engineering.
- Track record of publications in AI/ML for healthcare or neuroscience.
- Experience scaling experiments on cloud platforms (AWS, GCP, Azure) or HPC clusters.
- Knowledge of self-supervised learning, domain adaptation, and federated learning for cross-site generalization.
- Familiarity with ICU-specific challenges (e.g., predicting clinical deterioration, sepsis, ventilator weaning) or ALS research tasks (e.g., progression modeling, survival prediction, multimodal integration of speech, EMR, and imaging).
Responsibilities
- Develop and fine-tune LLMs (LoRA/QLoRA) for ICU/ALS note classification, temporal phenotyping, summarization, and structured JSON extraction (e.g., ventilator settings, ALSFRS-R scores, disease trajectories).
- Build retrieval-augmented generation (RAG) pipelines (hybrid retrieval, citation enforcement) with safety guardrails for generating evidence-grounded outputs from ICU/ALS corpora.
- Mitigate hallucinations and rigorously evaluate robustness, calibration, and fairness across ICU subpopulations (age, sex, comorbidities) and ALS cohorts (site, disease stage, language).
- Deliver reproducible pipelines with versioned data, containers, unit tests, and transparent evaluation metrics.
- Apply explainable AI (XAI) (concept attribution, counterfactuals, clinician-readable rationales) to enhance model interpretability in ICU monitoring and ALS progression modeling.
- Collaborate with intensivists, neurologists, and data scientists to co-develop models aligned with real-world ICU workflows and ALS clinical research.
- Contribute to manuscripts, grant proposals, and dissemination of findings at leading conferences and journals.
Other
- Ph.D. in Computer Science, Biomedical Engineering, Computational Neuroscience, Applied Mathematics, or related field.
- Strong problem-solving skills, independence, and collaborative mindset.
- Excellent communication skills for both technical and clinical audiences.
- Ability to work onsite at 101 Merrimac Street.
- Must be eligible to work in the United States.
