Adriana-Stefania Ciupeanu

I develop Bayesian, machine-learning, and network-based methods for complex biological and dynamical systems. My work connects mechanistic models with real-world data, with a focus on uncertainty quantification, probabilistic forecasting, and research software that other scientists can reproduce and extend.

As a postdoctoral fellow at the University of Manitoba, I lead the machine-learning extension of a global metapopulation model spanning 238 countries and territories. I designed a mechanistically informed neural-surrogate workflow for repeated model evaluation, likelihood-free Bayesian inference, and pathogen-importation forecasting. The end-to-end Python pipeline covers simulation, parameter-space sampling, PyTorch model training, SMC-ABC inference, posterior uncertainty propagation, validation, and computational benchmarking.

My earlier research includes COVID-19 variant dynamics, air-travel network change, HPV vaccination strategy evaluation, and analyses developed with Alberta Health Services and the Public Health Agency of Canada. I also build open-source Python tools for temporal network analysis and global transportation-network construction.

View research projects Download CV

What I bring

• Statistical modelling: Bayesian MCMC, SMC-ABC, parameter estimation, identifiability, sensitivity analysis, and uncertainty quantification.
• Scientific machine learning: neural surrogates, graph attention networks, recurrent models, PyTorch, and PyTorch Geometric.
• Complex systems: mechanistic ODE models, temporal and spatial networks, metapopulation models, and heterogeneous observational data.
• Research software: tested Python packages, reproducible computational pipelines, documentation, Git, and open-source collaboration.

I am interested in research scientist, statistician, quantitative researcher, and scientific software roles where careful modelling and reliable computation inform real decisions. I am based in Manitoba, Canada.