Postdoctoral Research Associate I - V (Center for Biomedical Informatics and Biostatistics (CB2))

• Develop novel ML / AI models and techniques for analysis of longitudinal time series data, medical images, patient logs and other relevant health related data points.
• Assist in managing study portals, data pipelines and visualization dashboards.
• Assist in maintaining production systems currently supporting various health science projects across campus.
• Participate in manuscript writing, grant preparations and presentation posters.
• Mentor undergraduate research interns or students.

Knowledge, Skills, and Abilities:

• Strong background in analysis of time series data including - Frequency domain decomposition (FFT), Wavelets, Forecasting Techniques, latent class modeling for time series data, visualization and exploratory analysis (seasonality, etc).
• A strong grasp on model evaluation metrics, model evaluation and training methodologies.
• An understanding of probability theory and basic frequentist statistical approaches.
• A fundamental understanding of Deep Neural Networks as applied to high-frequency time series datasets, including the ability to design and implement custom NN models in PyTorch, as well as the ability to implement custom loss functions, connections, and transformations.
• An understanding of class biases, model explainability (ex: Shaply), data augmentation techniques, missing data handling techniques, validation methods.
• Ability to analyze time series data including, Frequency domain decomposition (FFT), Wavelets, Forecasting techniques, Latent class modeling, Visualization and exploratory analysis (e.g., seasonality).
• A fundamental understanding of Deep Neural Networks as applied to high-frequency time series datasets, including: Designing and implementing custom NN models in PyTorch.
• Implementing custom loss functions, connections, and transformations Experience with LSTM, CNN, RNN, Encoder-Decoder, Transformer-based approaches Classification and regression techniques using NN models.
• Understanding of class biases, model explainability (e.g., SHAP), data augmentation, missing data handling, validation methods Strong grasp on model evaluation metrics, model evaluation, and training methodologies.
• Understanding of integrating Bayesian approaches in NN-based model
• Knowledge of model deployments to cloud platforms or past work with AutoML tools.
• Knowledge of MLFlow for maintaining model versions, datasets, evaluation metrics, and visualizations.
• DevOps (Optional) - Cloud (AWS / Azure / GCP / IBM), AutoML tools, Docker, AWS Athena, AWS Lambda, AWS SageMaker.
• Reproducible Research - Notebooks (Jupyter / Marimo), Git (or equivalent VCS).
• Programming Languages - Python, R (optional).
• Frameworks - PyTorch, TensorFlow (optional), MLFlow (optional).
• Databases - SQL (including queries using windows, joins, JSON operations.

• PhD degree in computer science, informatics or related discipline such as applied mathematics.
• Two (2) years of experience with using the following neural network model architectures: LSTM, CNN, RNN, Encoder-Decoder-based approaches, and Transformer-based approaches for time series data.
• Demonstrated experience with classification and regression techniques using neural network models, with at least 2 years of hands-on application.
• Experience with reading technical manuscripts.
• Three (3) years of experience writing clear and effective code documentation.

• Experience with reading technical manuscripts and writing code documentation.
• Experience of using analytical techniques in the health care space with data sparsity and class imbalances.
• A fundamental understanding of classical Machine Learning Techniques for longitudinal data analysis.
• An understanding of probability theory and basic frequentist statistical approaches.
• An understanding of integrating Bayesian approaches in NN based models.
• Experience of having using analytical techniques in health care space with data sparsity and class imbalances.
• Experience with model deployments to cloud platforms or past work with AutoML tools.
• Experience with MLFlow for maintaining model versions, data sets, evaluations metrics and visualizations.
• A fundamental understanding of classical Machine Learning Techniques for longitudinal data analysis.
• An understanding of class biases, model explainability (ex: Shaply), data augmentation techniques, missing data handling techniques, validation methods.
• Programing Languages - Python, R (optional).
• Databases - SQL (including queries using windows, joins, JSON operations).
• Reproducible Research - Notebooks (Jupyter / Marimo), Git (or equivalent VCS).

If invited to interview, please be prepared to provide three (3) professional references.

Please note, your cover letter should include a statement addressing how your research background aligns with the Preferred Qualifications and KSAs required for this position.