Max W. Shen

Hi, I am a Ph.D. Candidate at MIT. My research uses applied machine learning and statistical methods for fundamental scientific discovery and high-impact applications.

maxwshen [at]
Google Scholar
Cambridge, MA, USA

As a computationalist at heart, I believe in interdisciplinarity, getting one's feet wet, and that better solutions to real-world problems arise by marrying each problem's unique structure with thoughtful modeling and inference design. Some interests include genome editing, directed evolution, and the intersection between deep learning, causal inference, and Bayesian models.
I will soon complete my Ph.D. and am actively looking for opportunities for late 2020 or early 2021. I have published co-first author papers in Nature and Cell.
During my Ph.D., I worked with David R. Liu and Aviv Regev at the Broad Institute from 2018-now. Before that, I did my Ph.D work with David Gifford in the MIT Computer Science & Artificial Intelligence Laboratory from 2016-2018 while collaborating closely with Richard I. Sherwood. Before that, I began my Ph.D. at MIT in Computational & Systems Biology in 2015 after graduating summa cum laude with a B.S. in Computer Science with a specialization in bioinformatics from U.C. San Diego, where I did research with Pavel A. Pevzner and Mingxiong Huang.


Determinants of Base Editing Outcomes from Target Library Analysis and Machine Learning
Max W Shen*, Mandana Arbab*, Beverly Mok, Christopher Wilson, Żaneta Matuszek, Christopher A. Cassa, David R. Liu
Cell, 2020. My artwork was featured on the cover of the July 23, 2020 issue!
[Co-first author reordering approved by all co-first authors]
[Code] [Interactive web app]

Predictable and precise template-free CRISPR editing of pathogenic variants
Max W Shen*, Mandana Arbab*, Jonathan Y Hsu, Daniel Worstell, Sannie J Culbertson, Olga Krabbe, Christopher A Cassa, David R Liu, David K Gifford, Richard I Sherwood
Nature, 2018
[Code] [Interactive web app] [Press feature by Dash plotly for data visualization]

Continuous evolution of SpCas9 variants compatible with non-G PAMs
Shannon M Miller*, Tina Wang*, Peyton B Randolph, Mandana Arbab, Max W Shen, Tony P Huang, Zaneta Matuszek, Gregory A Newby, Holly A Rees, David R Liu
Nature Biotechnology, 2020

Assembly of long error-prone reads using de Bruijn graphs
Yu Lin*, Jeffrey Yuan*, Mikhail Kolmogorov, Max W Shen, Mark Chaisson, Pavel A Pevzner
Proceedings of the National Academy of Sciences, 2016

plasmidSPAdes: assembling plasmids from whole genome sequencing data
Dmitry Antipov, Nolan Hartwick, Max W Shen, Mikahil Raiko, Alla Lapidus, Pavel A. Pevzner
Bioinformatics, 2016

MEG source imaging method using fast L1 minimum-norm and its applications to signals with brain noise and human resting-state source amplitude images
Ming-Xiong Huang, Charles W Huang, Ashley Robb, AnneMarie Angeles, Sharon L Nichols, Dewleen G Baker, Tao Song, Deborah L Harrington, Rebecca J Theilmann, Ramesh Srinivasan, David Heister, Mithun Diwakar, Jose M Canive, J Christopher Edgar, Yu-Han Chen, Zhengwei Ji, Max W Shen, Fady El-Gabalawy, Michael Levy, Robert McLay, Jennifer Webb-Murphy, Thomas T Liu, Angela Drake, Roland R Lee
Neuroimage, 2014


Causal Inference & Deep Learning
MIT independent activites period, Jan. 2018
Max W. Shen, Fredrik Johansson
○ Prepared and co-taught a short graduate-level class with 4 sessions and 6 total h. Typical attendance: 20 students.

Applied Probabilistic Programming & Bayesian Machine Learning
MIT independent activites period, Jan. 2017
Max W. Shen, Alvin Shi, Carles Boix
○ Prepared and co-taught a short upper-division class with 6 sessions and 9 total h. First class attendance: 100 students, typical attendance: 25 students.


Applied Machine Learning: My work customizes powerful, modern deep models to leverage the unique structure within each real world problem. I have designed deep conditional autoregressive models to model base editing outcomes, and jointly-trained multitask sister deep networks to accurately learn a particularly noisy subset of CRISPR editing activity. I have taught classes to MIT undergrad and graduate students on Bayesian modeling, deep learning, and causal inference. I am proficient in pytorch, and enjoy keeping up with modern toolkits: see my low-level integration between pyro and gpytorch to learn 10 latent Gaussian processes on time series data in a larger probabilistic model with stochastic variational inference.

Software Engineering: See my GitHub. I completed software engineering internships at Qualcomm Korea (2013) and Illumina (2014). Python is my language of choice, though I have previously worked in C++. Summa cum laude B.S. in Computer Science with a specialization in bioinformatics (2011-2015).

Data Visualization: My interactive web apps have received press attention from Dash plotly. I am proficient in Adobe Illustrator, Photoshop, Premiere Pro, and After Effects, and use matplotlib, pandas, and seaborn everyday. I am also proficient in html, css, and dash plotly.

Communication and Collaboration: My Ph.D. has featured extensive collaboration with wet lab experimentalists, including Richard I. Sherwood and Mandana Arbab. I have completed a 40 h course on conflict management and mediation that has substantially impacted my life.

Management: For two years in undergrad, I managed 9 teams with a total of ~100 students to host a regional urban-style (hip hop) dance competition with ~2,000 audience members with revenue and expenses of $35,000/year. Each team had ~10 students and 2 team leaders, and all team leaders were overseen by me and one other co-leader. This provided substantial public speaking experience, as I led dozens of meetings speaking to and motivating our team of 100 students.

Max W. Shen is a Ph.D. candidate at MIT, working on applied machine learning and statistical methods for fundamental scientific discovery and high-impact applications.

This website was designed with Skeleton CSS.