Prospective Lab Members

Thanks for your interest in working with me! Please read this page before contacting me.

In general, I plan to hire 1-2 PhD students per year, but that may fluctuate year to year. Below I describe some of the attributes I look for in prospective students, and some of my philosophy on what makes for a great PhD and a great advisor-student relationship.

Technical Qualifications

The most important attributes of a successful PhD student are being highly self-motivated, persistent, and curious. I expect students to take ownership of their PhD research, choose projects they are excited about, and see those projects through to completion. Some specific technical background that I expect from students:

• Mathematical maturity. I expect students to have – or build – a solid foundation in linear algebra, optimization, signal processing, basic probability & statistics, and basic algorithms & data structures. Students should be comfortable reading (and ultimately writing) papers that involve equations and/or algorithm pseudocode.
• Coding proficiency. I expect students to be comfortable building projects from scratch, as well as building on existing code. I expect most of our projects to use Python (including GPU and autodiff libraries like PyTorch, JAX, and CuPy), but if you are familiar with a lower-level language like CUDA or C++ that is a bonus. I expect students to have – or build – basic familiarity with version control (git) and ssh command-line tools (e.g. tmux). Lab members are expected to maintain version history on our lab GitHub page, and write code that is clean, modular, and well-documented to allow others (and our future selves) to build on our work.
• Technical communication. I expect students to be able to clearly communicate what they did in their project and why it matters – both in technical writing and in formal and informal oral presentations. A good place to start when “pitching” a new project idea – to me, to a colleague, or in the introduction of your paper – is by answering the first few questions in the Heilmeier catechism.

You don’t need to be an expert in all 3 categories to start your PhD, but you should have basic competence in all 3, strength in at least one of the first 2, and interest in building expertise in all 3.

What to Look for in an Advisor-Student Relationship

• Mutual respect. My job is to help you develop as an independent researcher. Mutual respect includes respecting each other’s time and research interests, and being open to feedback.
• What to expect. At the beginning of the PhD, the relationship is somewhat like an apprenticeship – you might start working on projects I suggest, and/or help a more senior student or postdoc with their project. Towards the end of the PhD, the relationship is more like colleagues – I expect you to define your own research agenda and pitch project ideas to me rather than the other way around.
• Prioritize finding an advisor you work well with. Consider the questions in this checklist as you search for a PhD advisor. A good fit includes both sufficient alignment/overlap in research interests and compatible advising style and communication style.
• Publication expectations. I prioritize high-quality and impactful research, and sometimes that means finding a balance between finishing a project to make a deadline vs. deciding to wait and publish a more comprehensive paper later. A rough target to aim for (at least towards the beginning of your PhD) is one first-author publication per year, though research progress is often nonlinear and you will likely be more productive towards the end of your PhD.

If you think I might be a good advisor (or postdoc mentor) fit for you, please email me and include the word “voila” as the first word of the subject line of your email. In the body of the email, please attach your CV and mention your background/experience, research interests, and why you are interested in joining my lab (e.g. are there particular papers I’ve written that you find compelling?). You don’t need to have any published papers to join my lab as a PhD student, but please do describe any research you have taken part in (published or unpublished), what your role was in the project(s), and what you learned from the experience (both technically and about yourself and your research style/preferences/strengths).

If you are a prospective student not yet at Georgia Tech, please apply to the ECE and/or ML PhD programs and mention my name as a potential advisor in your application. Please also consider applying to any fellowship programs you may be eligible for, including NSF graduate and postdoctoral fellowships (for US citizens), industry graduate fellowships, and (for prospective postdocs) the President’s Postdoctoral Fellowship at Georgia Tech.

Advice for New PhD Students

• Be persistent. Research takes time, and usually you’ll need to try several ideas that don’t work before you find one that does. You’ll need to commit sufficient time to your research for it to make progress (treat research as your full-time job, especially as your course load decreases).
• Be optimistic. It feels awesome when you have an idea that works. Know that if you are persistent you will get there!
• Be curious. Often in the course of working on a research project you will encounter something unexpected. It might be a bug in your code, or it might be a new phenomenon you have just discovered. Pay attention to weird things, and try to figure out what’s going on. Be open to changing the course of a project, trying something you didn’t originally plan on, and learning new topics/skills as you they become relevant to your research.
• Communicate effectively. Plan ahead for meetings, to use time efficiently. If I made suggestions in our previous meetings, have you implemented them? Describe what you did on research since our last meeting – using a mixture of figures, equations, and/or pseudocode, with enough detail that I can interpret your results. What worked and what didn’t work? Do you see any patterns? Did anything unexpected happen? What do you think is the next thing to try? Is there anything blocking you (if you are blocked with several days until our next meeting, message me on slack instead of waiting for the meeting)? Do you have any higher-level (e.g. career advice) questions for me?
• Own your PhD. I am here to guide you in directions I hope will be fruitful, but ultimately it is your PhD and your responsibility to take ownership of your projects and see them through to completion (which usually means publication, code release, etc.). If there’s a direction you want to go, let me know!

I am compiling a list of resources, advice, and expectations for current lab members in this VOILA lab institutional memory document.