I have three main areas that I'm excited about: probabilistic models, reinforcement learning, and interpretable machine learning. All of my work currently focuses on questions drawn from realworld technical challenges associated with using machine learning to assist in healthcare contexts. If you think that I'd be a good mentor, please get in touch after reading the FAQs below.
.
Are you taking undergraduate students? In general, I only have the capacity right now to work with students from Harvard and local collaborations such as MIT and Wellesley. Experience with some kind of numerical computing (e.g. Matlab/numpy/scikit-learn, at the level of CS181) is a must, and additional statistics will allow for more interesting projects. In most cases, I ask for a two-term or thesis level of commitment, and I have a strong preference for several students working together on a joint project aimed toward a publication or product. Usually the best time to reach out is in the spring for the following fall. I only advise students on projects that are related to the core directions of my lab -- it really isn't fair to anyone if I don't have the brainspace to think about your project! -- but students are always welcome to come by at office hours to discuss ideas.
.
Are you taking PhD students? I expect my graduate students to already have taken some courses in probability/statistics, linear algebra, and numerical computing -- and be excited about seeing a lot more of these areas during their PhD! I've found that applying models to real health and wellness settings makes one care a lot more about the basic properties of models, optimization, and communication, so we do a lot of basic research in synergy with applied work. I'm also expanding into HCI-related areas in collaboration with my HCI colleagues. Given my current lab size, I am highly unlikely to be taking graduate students for Fall 2024.
Note that while you must be accepted to Harvard SEAS for me to consider you in the group, it may be useful for you to drop me a note also in advance. If you get in touch before applying, please share your CV, transcript, and any other relevant experience (e.g. if you've coded up your inference for a class or project) is helpful for me to get a better sense of your technical background.
Finally, there are some areas that I do not work in -- imaging, NLP, omics, atari -- if you have experience in these areas, that totally counts as research cred; but if you're hoping to do more of that in your PhD, I'm probably not the lab for you.
.
Are you taking Masters students? I do not take Master's students directly, nor do I have particular sway in admissions to the data science master's program (where my role is most associated with curriculum and staffing). If you are admitted into the master's programs and want to chat, please reach out then.
.
Are you taking postdocs or interns? Unfortunately, I do not currently have the capacity to hire short-term postdocs or interns. If you are looking for a longer-term postdoc position (e.g. 2 years) and think that I might be a good match, please get in touch! I expect postdocs to have completed their PhD in machine learning, math, stats, physics, or some other technical area and already have deep experience in inference (can be sampling, optimization, etc.); you will have the chance to mentor both undergraduate and graduate students in these areas (as it relates to joint projects).
.
What other skills do I need? I expect all of my students to have excellent communication skills and the ability to write software independently (i.e. I'm always happy to help, but I'm not going to be teaching you how to code). More than any other specific skills, however, I'm looking for passion and grit to learn from failures and persevere when tackling tough, meaningful problems. If you're reaching out to me, feel fee to include a personal story of working through something really exciting and really hard, which may or may not have worked out :)
.
What are your collaboration expectations? I expect my students to be organized and timely with (at least) weekly updates. I also expect my students to communicate challenges so we can work through them together.
.
Anything else before getting in touch? Please show me that you've read the text above by structuring your subject line as "Purple Unicorn: [ FILL IN YOUR SUBJECT LINE ]"