Wei Li

About me

Ph.D. in Statistics · Focus: Machine Learning, Causal Inference, Deep Learning

Interdisciplinary training in statistics, econometrics, and mathematics, with research experience in Bayesian inference, statistical inference, causal inference, and teaching experience in statistical learning and computational methods.

Recent work has expanded into deep learning and generative models, with an interest in interpretable and causal approaches. Holder of the IBM AI Engineering Professional Certificate.

Selected Projects

Statistical Learning

Core concepts and methods in statistical machine learning, including regularization, nonparametric methods, ensemble methods, bootstrap methods, tree-based models, and neural networks, with model evaluation and statistical inference.

Statistical Computation and Simulation

Optimization and simulation-based techniques, including Monte Carlo, EM algorithm, Metropolis–Hastings, Gibbs sampling, slice sampling, simulated annealing and other numerical algorithms.

Deep Learning (PyTorch)

Developed neural network models, including convolutional and recurrent architectures, attention mechanisms, transformers, and generative models (GAN), with structured training and evaluation workflows.

Deep Learning (NumPy)

Built from-scratch neural network architectures to study implementation, including multilayer perceptrons, 1D and 2D convolutional neural networks, and recurrent neural networks. Results are benchmarked against PyTorch implementations to ensure accuracy and performance validity.

Causal Inference with Deep Learning (in progress)

Development of causal machine learning methods for treatment effect estimation, matching, propensity score weighting, regression adjustment, doubly robust methods, TMLE, DML, LATE, CATE (S-learner, T-learner, TARNet, CFRNet, DragonNet, X-learner, R-learner), causal forests, GRF, Bayesian causal forests.

Applied Skills

Machine Learning & AI
Supervised and unsupervised learning, deep learning, causal machine learning, generative models

Statistical Methods
Bayesian inference, Causal inference, nonparametric and semiparametric methods, Monte Carlo and MCMC

Programming & Frameworks
Python (NumPy, Pandas), R, SQL; PyTorch, TensorFlow, Keras, LLM prompt programming (DSPY)

Educaiton

Ph.D. in Statistics, North Carolina State University, 2018
M.A. in Mathematics, Indiana University Bloomington
M.A. in Economics, George Mason University

Professional Experience

Assistant Professor, Syracuse University (2018–Present)
Research and teaching in Bayesian inference, causal estimation, and computational methods, with focus on modeling and inference.

Econometrics R&D Intern, SAS Institute (2017, Summer)
Developed and validated algorithmic components for Hidden Markov Models within the SAS Econometrics platform.

Selected Publications

• Li, W. and Ghosal, S. (2021). Posterior contraction and credible regions for level sets. Electronic Journal of Statistics.
  
• Liu, J. and Li, W. (2021). A semiparametric method for evaluating causal effects with error-prone covariates. Biometrical Journal.
  
• Escanciano, J. C. and Li, W. (2021). Optimal linear instrumental variable approximations. Journal of Econometrics.
  
• Li, W. and Ghosal, S. (2020). Posterior contraction and credible sets for filaments of regression functions. Electronic Journal of Statistics.
  
• Li, W., Yang, S., and Han, P. (2020). Robust estimation for moment condition models with missing data. Journal of Statistical Planning and Inference.

Technical Focus

Causal machine learning, deep learning, large language models, and reliable and interpretable AI systems.

Resume/CV available — contact via email .