Biyonka Liang

Harvard University / Department of Statistics

About Me

I am a Ph.D. Candidate in Statistics at Harvard University advised by Professor Lucas Janson in the Department of Statistics and Professor Iavor Bojinov in the Harvard Business School. My research focuses on developing statistical methods and models for challenging practical problems in adaptive experimentation, reinforcement learning, and causal inference. I am gratefully supported by the NSF Graduate Research Fellowship. Prior to joining Harvard, I was a Data Scientist at Meta. I received my Bachelor degree in Statistics from UC Berkeley, where I was awarded the 2019 Departmental Citation as the top graduating student in the statistics bachelor's program. During my Ph.D., I was a Biostatistics Intern at Moderna working on clinical trial planning, a Research Intern at Zuse Institute Berlin working on combinatorial optimization, and a Statistical Consultant at MDRC working on data-driven policy evaluation.

Working Papers and Publications

A Bayesian Approach to Online Learning for Contextual Restless Bandits with Applications to Public Health

Liang, B., Xu, L., Taneja, A., Tambe, M., Janson, L. (2024)

arXiv

An Experimental Design for Anytime-Valid Causal Inference on Multi-Armed Bandits

Liang, B., Bojinov, I. (2024)

arXiv

Powerful Partial Conjunction Hypothesis Testing via Conditioning

Liang, B., Zhang, L., Janson, L. (2023)

Under Review, Biometrika.

arXiv

Vector bionomics and vectorial capacity as emergent properties of mosquito behaviors and ecology

Wu, S., Sánchez, H., Henry, J., Citron, D., Zhang, Q., Compton, K., Liang, B., et al. (2020)

PLoS Computational Biology, Volume 16(4), e1007446

DOI

Conference Presentations and Talks

"An Experimental Design for Anytime-Valid Causal Inference on Multi-Armed Bandits". Invited talk for the session "The Design and Analysis of Complex Experiments" at the 2024 American Causal Inference Conference in Seattle, USA.
"An Experimental Design for Anytime-Valid Causal Inference on Multi-Armed Bandits". Talk at the 2023 Conference on Digital Experimentation @ MIT in Cambridge, USA.
"Powerful Partial Conjunction Hypothesis Testing via Conditioning". Poster presentation at the 2023 Joint Statistical Meetings in Toronto, CA.
"Network Analysis of Mosquito Habitats for Controlling Vector-Borne Pathogens". Poster presentation at the 2018 International Conference for Network Science in Paris, FR.
"MICRO: An Eco-epidemiological Agent-based Framework for the Modeling of Mosquito-borne Pathogens". Talk at the 2017 NorCal CompBio Symposium in Santa Cruz, USA.

Honors and Awards

American Statistical Association Gertrude M. Cox Scholarship, Honorable Mention (2022)
Datathon at the Laboratory for Innovation Sciences at Harvard, Top 10 Finisher (2022)
Harvard University Certificate of Distinction and Excellence in Teaching (2021, 2022)
National Science Foundation Graduate Research Fellowship Recipient (2020)
Citadel Data Open, Summer Invitational, First Place (2019)

Teaching

Stat170: Stochastic Processes, Harvard University (2022)
Stat211: Statistical Inference (graduate-level), Harvard University (2021)
Stat140: Probability for Data Science, University of California, Berkeley (2018-2019)
Stat134: Concepts of Probability, University of California, Berkeley (2018)

Contact

Email: biyonka@g.harvard.edu
1 Oxford St, Cambridge, MA 02143
Full CV Google Scholar Github

Misc

Outside of academics, I love to figure skate, ski, and sing. I was fortunate to be a member of the Radcliffe Choral Society from 2020-21 and am an active member of the Harvard Figure Skating Club. I am a part of the amazing team of faculty, students, and staff who helped found Harvard's Data Adventure Day, an event designed to promote statistics and data science to high school student through hands-on activities and interactions, with a focus on supporting diversity and inclusion in these fields.