CS Ph.D. Information Session — 18 November 2020

We held our first CS Ph.D. information session last Wednesday for 285 registered attendees. Below, you can find links to questions we answered on YouTube and written answers to the questions we didn’t get to. Also, there’s another session coming up on December 3rd.

The panel of professors and graduate students

Links to Video

QuestionDirect Link
What are some of the qualities of an ideal candidate for a PhD? Conversely, what might raise a red flag in a candidate’s application?https://youtu.be/thZdtIQ3i_g?t=934
How do you find a topic and an advisor?https://youtu.be/thZdtIQ3i_g?t=1392
What matters most in the application? Undergrad grades, graduate (MS) grades, GMAT/GRE, recommendations or research experience?https://youtu.be/thZdtIQ3i_g?t=1123
How does work experience as software engineer factor in? Or does that only count if you’re part of a published research project?https://youtu.be/thZdtIQ3i_g?t=1696
What job opportunities in industry does a PhD or Master’s in CS open and close compared to a Bachelor’s?https://youtu.be/thZdtIQ3i_g?t=2752
Is there a time of life that is best to pursue a PhD? Is there an average age for a first year PhD candidate?https://youtu.be/thZdtIQ3i_g?t=2954
What kind of job can I get with a Ph.D. and what’s the job market like?https://youtu.be/thZdtIQ3i_g?t=1827
Can you do a Ph.D. program part time?https://youtu.be/thZdtIQ3i_g?t=2495
Can I have good work life balance as a Ph.D. student?https://youtu.be/thZdtIQ3i_g?t=1965
How important is it to reach out to potential advisors before applying? Can they vouch for you in the application process later on?https://youtu.be/thZdtIQ3i_g?t=3163
Should I still apply if I don’t have any research experience?https://youtu.be/thZdtIQ3i_g?t=2250
How do you pay for a Ph.D. and how much does it cost?https://youtu.be/thZdtIQ3i_g?t=778
What’s the weight of different parts of the application (test scores, grades, letters, SOP, undergraduate research)?https://youtu.be/thZdtIQ3i_g?t=934
Do I need to get as master’s degree before getting a Ph.D.?https://youtu.be/thZdtIQ3i_g?t=778

More Questions and Answers

How do you get research experience late in your undergrad?

Steven Swanson — Professor

It depends on how late. If you have a summer left, you could apply to an NSF REU program. Asking professors for classes that you enjoyed and did well in is another route. Finally, you can think broadly about what “research experience” means for the purposes of grad school applications. What we are really looking for is evidence of your ability to make progress on large-scale, open-ended problems and projects. So joining (and being active in) something like a video game dev club, or taking on an ambitious independent study could both help.

Akshat Singhal — Grad Student

If you are already in your senior year and haven’t applied yet, consider joining a research lab. Lots of research labs have job openings where you can work for a year or two after graduating. Not only will it provide you with research experience, but it will also tell you whether you like research or not.

Sachiko Matsumoto — Grad Student

Joining a research lab at your school can be a good opportunity. You can often get units for doing research (e.g., you can sort of count it as a class). Some students stick in one lab during their undergrad, and other students switch partway through to get experience in different areas/see what they like.

 

What do REU programs look for in an applicant?

Steven Swanson — Professor

Here’s an answer from the REU at UCSD (Run by Ryan Kastner): “I’m guessing that our program is unique in this regard, but we focus almost exclusively on system building experience (e.g., robotics clubs they are involved in, competitions, unique classes). Some level of research experience helps, but most of our applicants don’t have that. We also look favorably on leadership roles and evidence of being able to function well in a team.”

 

What are the benefits of a CS PhD vs a CS Masters Degree?

Steven Swanson — Professor

The benefits are that you get trained to do research-style work rather than “just” gaining more depth of knowledge in CS. It varies, but MS degrees are largely about taking more advanced classes for a couple of years. That’s great, and you will learn a lot. A Ph.D. includes that but also adds in a completely different set of skills around identifying hard problems and devising new solutions to them. It opens up some additional career opportunities in areas like advanced R&D and industrial research and, of course, academia.

Mark Schultz — Grad Student

A pro/con is funding. PhDs tend to be funded, which means they are generally more competitive to get into. The funding is usually either for TAing or research, so while you’re in the degree you’ll have TAing or research obligations. But you get money, which is nice. Also at UCSD PhD students are “treated better” by the bureaucracy of things — if there are limited seats in a class, PhD students get first dibs. Same for TAing opportunities.

 

I don’t have a very strong GPA(3.1), however, I have a NeurIPS workshop publication and other internship experiences. will my GPA affect the admissions?

Steven Swanson — Professor

Yes, but the NeurIPS paper will help. I would not assume that either will sink or save your application, so apply widely.

 

How do graduate courses compare with undergraduate courses, in terms of difficulty and workload?

Steven Swanson — Professor

From the instructor’s perspective, they are less structured, more abstract, and move a good bit faster. The workload varies, but in my experience (and in my classes) the projects are somewhat smaller because Ph.D. students should be spending a lot of time on research and many professors recognize that their research should be their top priority.

Mark Schultz — Grad Student

In my experience, they are much easier. This is both because the coursework requirements are usually lesser, you take fewer of them at a time, and because they tend to be closer to something you’re already familiar with (so often take less grappling with entirely unfamiliar concepts).

Sachiko Matsumoto — Grad Student

In my experience, they tend to be easier, but it can vary a lot from class to class and depends on your prior experience/classes.

 

Studies showed that Ph.D. students face mental health challenges that sometimes lead to suicide. What policies does UCSD have to deal with this issue?

Steven Swanson — Professor

That’s an excellent question and one that deserves more space than we have here. The short, UCSD-specific answer is CAPS: https://wellness.ucsd.edu/caps/Pages/default.aspx

Mark Schultz — Grad Student

Something I’ve really valued that UCSD does is cultivating a good departmental culture. During non-pandemic times there are weekly events that encourage students in different areas to hang out (usually by bribing us with free food/coffee), which makes everything feel more open and welcoming. The location also helps a bunch — both due to being closer to the equator (which helps prevent seasonal affectiveness disorder), and just because the weather is just ridiculously good.

 

Very specific question — do people look at your Github during the application process?

Steven Swanson — Professor

We will look at anything you highlight in your application, but I’m not going to google you on github.

 

What jobs help with grad school apps if we are taking a gap year?

Steven Swanson — Professor

I would say either something technical and related to the area you want to do research in.

 

Does the PhD selection process in UCSD involve selection via an admissions committee or through professors you put down in your application?

Steven Swanson — Professor

It varies by school but it is usually a combination. At UCSD, the admission committee filters out unqualified students but doesn’t actually admit anyone. The faculty look at the qualified students and have them admitted.

 

Have any of you had to pivot your research direction midway through the PhD — how did it go?

Nadia Heninger — Associate Professor

I completely changed research areas during grad school, from algorithms/theory to applied cryptography and security. It took me a couple of years to accept that I was having much more success with the research that I thought was just “side projects” than the topics I thought I wanted to be my actual focus. I ended up with several mentors outside of the university, and I also did a couple of years of postdoc after grad school to get a stronger publication record before applying to faculty positions.

Akshat Singhal — Grad Student

These cases are rare and I personally didn’t have to make any changes but I know two people who had to.
A friend of mine had to change her advisor in her third year of PhD. Her previous advisor suddenly asked her to leave the lab. However, because PhD students are technically hired by the departments, she didn’t have to quit. She was able to find an advisor after a qaurter. Her current advisor doesn’t have funds to support her and so she is TAing to get the stipend. It’s definitely not what she expected but her program had a fail-safe for such a case.
Another friend of mine had to look for a new advisor in the first quarter itself because their advisor moved to another school. Through coursework, journal clubs, and papers they realized they are more interested in a different field (moved from bioinformatics to fairness in AI/ML). It took them 2 quarters to find a new advisor but their current advisor is funding them and is also very kind (paraphrasing).

Sachiko Matsumoto — Grad Student

I don’t know of anyone who has completely changed their research area, but a lot of people I know change what they focus on in the first couple years of their Ph.D. For instance, I started on a trust/human robot teaming project my first year and am switching to more shared control stuff. One of my labmates started on an activity recognition project and then switched to dementia/MCI focused work.

 

If you already have a master’s degree, how much work is eliminated from the Ph.D. work?

Steven Swanson — Professor

The first 2 years of a Ph.D. are partly consumed with course work. Each program will have some courses you need to take. Your MS coursework can eliminate some (or even most) of this, but if there’s a mismatch between your MS course at your MS school and the Ph.D. requirements at your Ph.D. school you may have to fill the gaps.

Akshat Singhal — Grad Student

Theoretically, you can get all your coursework requirements waived off. In that case, you’ll get your PhD with a 0.0 GPA. I took 6 courses during my master’s and I got 5 of them waived off. At UCSD, we have to take a minimum of 9 courses, so I only had to take 4 courses.

 

What advice can you give to current juniors to help prepare for PhD in CS?

Mark Schultz — Grad Student

Things that would help any PhD application are having good letter writers or research experience (generally these go together — the best letter writers are those who have experienced working with you on research). Making a goal of having both by next year would be a good idea. Note that “research” does not have to be super elaborate — if a professor at your home institution hands you a few papers, has you read through them and talk about them and is impressed by your understanding of the papers, this is already a good start on having a good letter writer.

 

What do you usually do during the summers? Internships, working in the lab, etc?

Mark Schultz — Grad Student

I’ve mostly been working in with my advisor, but really should be doing (research hopefully) internships. Software engineering internships are an option as well, but in my field (theoretical cryptography) they are less common in my understanding.

Akshat Singhal — Grad Student

It depends very much on the advisor. Most advisors prefer that their students work in the lab as summers are a very good time to move the project forward because advisors usually have more time during summers. However, sometimes they may not have funds or there might be an industrial project that could help in your research. There are also these industrial fellowships where the students have to spend the summers at company headquarters.

 

What is the most challenging stage among other stages such as course requirements, research exam requirement, or the qualifying examination requirement?

Nadia Heninger — Associate Professor

The hardest part of a PhD is learning to do research. You’re trying to do something that nobody’s ever done before, and it might not work. The material in your undergraduate courses has been digested into lecture-sized pieces and translated into textbooks, and the assignments are designed to be doable. Research can involve a lot of failure and rejection and trying to puzzle through entirely new ideas. The course and exam requirements in a PhD are fairly straightforward hoops to jump through if the harder parts (making progress on research, learning to write papers, and staying motivated) are going ok.

What is the typical route a PhD student follows after earning a PhD? What if one’s goal is to teach at the college level?

Steven Swanson — Professor

If you want to teach in college, you will often apply for those positions during the last year of your Ph.D. and start the following fall. Otherwise, you’ll apply for (and hopefully get) some other job after graduating with your Ph.D.

Nadia Heninger — Associate Professor

Teaching at the college level can mean multiple things. If your goal is a tenure-track position at a research university, then your priority is strengthening your research record. Doing one or two years as a postdoctoral researcher at another university after your PhD is becoming more common in CS, though it is nowhere near required the way it is in some other scientific fields. Your advisors would help you figure out when you are ready to go on the academic job market, and you would apply for faculty or postdoc positions in the fall of your last year as a PhD student or postdoc, interview in the spring, and then hopefully have an offer to start the next year.

If your goal is a more teaching-oriented position at an undergraduate-focused institution or as a lecturer at a research university, then you would likely try to gain more solo teaching experience during your PhD by, for example, teaching some large classes. Then you would apply to faculty or lecturer positions in your last year as a PhD student. People who get jobs in industry often already have connections at some companies from doing summer internships during their PhDs. You would apply for industry positions after your PhD in a similar way to students coming out of undergrad, except that you have several years of experience in grad school and may have more connections.

 

How do you deal with the stress of a graduate program?

Mark Schultz — Grad Student

To a certain extent its like how you’d deal with the stress of any job, although maybe amplified a bit. Common suggestions would be finding a therapist, having a good support group generically (the other students in your grad program are a good resource, but having people outside academia can be good too), and having interests separate from your work

Akshat Singhal — Grad Student

Having a peer group is very important, people you can hang out with (not including the pandemic), vent off to when stressed. I make sure to have a weekly “me time”. I personally prefer to not work on the weekends (Friday night – Sunday noon). However, if I have to work on a weekend, I may take extra time off in the middle of the week. I play board games, tennis, read fantasy novels, watch movies/tv shows, do some volunteer work during this time. I also make sure to eat well and never compromise on my meals.
Almost equally important to all of this is having an understanding advisor. You should be able to talk to your advisor openly about any work-related stress (sometimes even personal). This is something you should definitely ask current grad students before joining a lab.

 

I saw a part-time question below, but want to reiterate that many, esp. those in ~security engineering can’t leave work for research. Best way to pursue this?

Mark Schultz — Grad Student

To pursue a part-time PhD? It is rather uncommon (and in fact may be disallowed by the UC system — I have not read our TA contract in-depth but have heard that there’s something in there disallowing outside income).

 

I’m interested in HCI research and it seems that CogSci has more HCI faculty. Does it increase my chance to be admitted if I apply to both?

Steven Swanson — Professor

Yes, it will. Admissions between departments are totally separate, although some professors have appointments in both.

 

If I work in industry for a couple of years, then go to a PhD, how is that viewed in the application process?

Steven Swanson — Professor

Not negatively and probably positively depending on what you did, where you worked, and how it appears to have informed your career goals and motivation for getting a Ph.D.

 

What is the efficient length of a SOP? Is a long SOP (say a 2 page) considered worse than a 1 page SOP (especially for UCSD PhD program)?

Steven Swanson — Professor

As with all writing, you should be as brief as possible while still conveying the necessary information. The most important thing to convey is why you are getting a Ph.D. and why you want to work with the particular faculty you want to work with. Stories about how you fell in love with the program when you were 12 (while pleasantly nostalgic) are not as useful.

 

Are all programs in person or is there opportunity to pursue a Ph.D. virtually or a mixture of both?

Steven Swanson — Professor

I would not plan on being able to do a Ph.D. remotely. The overwhelming common case is to do it in person. Getting a Ph.D. has a substantial cultural and social component. You are just learning about computer science, you are learning to be a researcher.

 

Does professors or the admission committee make the final decisions on admission? Or which side is more important?

Steven Swanson — Professor

It varies by school. At UCSD, one way to think about it is that the admission committee can say “no” but only a faculty member can say “yes”.

 

How are low impact publications vs high impact publications viewed in the process?

Steven Swanson — Professor

High impact pubs are better than low-impact pubs, but pubs at all for an undergrad are a good sign. In most cases the impact of the pub is not that important: except in very rare cases undergrads are not responsible for the things that make publications “high impact” (e.g., identifying or solving a really important problem). The publication mostly demonstrates that you can successfully contribute to a publishable research paper. If it’s a really good paper, so much the better, but key part is that you actually participated and contributed.

 

How long is a typical PhD program?

Steven Swanson — Professor

It varies by research area, my guess is 5-8 years.

 

If you are working in the field that you want to pursue your Ph.D. in, can you continue to work?

Steven Swanson — Professor

Doing a Ph.D. part-time is hard if not impossible. This applies even if your research area is related to your previous job.

 

What are you all currently researching? What’s something exciting you’ve learned thus far?

Akshat Singhal — Grad Student

My research area is bioinformatics. I’m currently working on interpretable deep learning models that predict drug response in cancer cells. I come from a more conventional CS background (algorithms, OS, android/web apps) so everything I read about is exciting.

 

If you don’t have a Bachelor’s degree in CS but another non-related field, what steps should you take to be competitive for a PhD in CS?

Nadia Heninger — Associate Professor

There are two considerations. Specific requirements vary by program, but in your first two years, you will need to take and pass a number of classes or exams spanning different areas of CS from theory to systems. You will need enough background in CS to get through these requirements, and to convince a potential advisor you can get through these requirements. Second, in your application, you should make it clear that you understand CS research and can articulate to your potential advisors why you want to do a PhD in CS.

Mark Schultz — Grad Student

It depends on how related. My undergrad was (essentially) in math, but I took a handful of related courses (cryptography, complexity theory, and an algorithms course that covered streaming/sketching algorithms). That kind of background is “typical” for someone interested in theoretical computer science, but I’m sure I would have been rejected if I was applying for an architecture PhD. You want to be able to make the case that you’ll be able to succeed. But for concrete things to do, I’ve heard of people doing a masters in the area first, and then trying to transfer/reapply to a PhD program.

 

In regards to speaking with an advisor’s current grad students, how would you recommend locating these individuals? Do labs have staff directories to check?

Mark Schultz — Grad Student

It is very common for professors to list their current and past students on their webpage. This is the first place I would check. Otherwise, you could try looking through their recent publications for coauthors from the same institution, but that sounds more annoying.