Lessons learned from Navigating and surviving my PhD defense -- and How that's helped in industry

November 2022

As you can probably tell from my name, I passed (!) – about 10 years ago now. I’ve learned many valuable lessons along the way, some of which took years to sink in, after seeing friends encountering similar challenges and post-mortem reflection.

Before we get started, I wanted to thank my PhD committee again, for supporting me as I navigated the uncharted waters of inventing a new methodology, then proving it out; thank you Amy, Hongquan, Jan, Kuo-Nan and Mark.

I wanted to get a PhD to work at JPL, to continue space exploration, which I started as a NASA Graduate Student Researcher. Things didn’t pan out that way; when I graduated with my PhD, NASA had budget cuts and was not hiring. I applied for teaching, post-doc and industry roles – industry got back to me first. While I haven’t used the exact methodology I invented for my PhD in industry, a PhD has helped me learn how to learn, which in turn, has helped me navigate the ever-evolving Tech world.

I’ve learned many lessons along the way! Whether you’re just starting your PhD program or getting ready to defend, here are my lessons learned.

  • Bonus: these are also applicable in industry!

In no particular order…

PhD Requirements

  • Find out from your committee what are all of their requirements and expectations that committee members have for you, so that you can get your PhD. Please note: there may be additional requests from specific committee members on these requirements, which may be above and beyond what your Department requires.

        • I highly encourage you to do this at least:

              • ASAP after you form your committee,

              • confirm again around orals, and

              • again (!) around the defense.

        • This way, there are no surprises! You don’t want to defend, then find out after (as have happened to a few of my friends following their successful defense), that in addition to inventing a new method, proving that it works and defending, they were expected to have a paper submitted/published to get their PhD.

        • Don’t forget to find out if your committee will be OK with you going into industry after getting your PhD?

              • At the time, I was one of a small number of PhD candidates in the history of the Department to go industry, rather than academia after graduating. It’s much more common to do so now.

        • In industry, I aim to figure out who the stakeholder(s) is/are that will be using my deliverable to help them make decisions and uncover the business question (including end goals and outcomes), before work begins, so my stakeholders and I are on the same page as early as possible.

    • Is there an approved department/university LaTeX/Word/etc. template that your university requires you use, when writing and submitting your dissertation? Who's that person at the university who signs off that your thesis followed the required formatting guidelines?

        • In industry, I check for company-branded templates, and tend to follow an outline something along these lines for presentations:

              • Statement of the Problem

              • Executive summary, including actionable findings/takeaways

              • (As needed ) Context/story to understand statement of problem, including any assumptions made

              • 1-2 proposed next steps to discuss

              • Appendix going into more detail on everything, including topics of further research

Communication

  • Check-in with your stakeholders more than once per year (!). I came up with my dissertation topic, proposed it to my advisor, and… my committee only heard from me when I proposed my topic and then again when I was defending – more than a year apart! In-between the two meetings, I pivoted applications of my methodology from understanding Earth’s atmosphere to Jupiter’. Understandably (in hindsight), as an expert on the work, I forgot to share the context for the pivot – and there were a lot of questions!

        • On my first call(s) with (industry) stakeholders, I aim to learn what check-in cadence works for them, to make sure we’re always on the same page; typically that’s every 1-2 weeks.

  • Can a non-technical, non-expert audience follow along with your defense presentation?

  • As you draft your presentation for your defense and your dissertation, don’t forget to ask for feedback/help! I know you’re doing that PhD for you, and you’re slowly becoming THE solo expert on your novel approach. That doesn’t mean that no one can help you be a sounding board on your presentation, debug your code, give advice on what to consider, etc. You have lots of resources at your disposal, including: Google/StackOverflow, your advisor, classmates, on-campus consulting center(s), on-campus writing center and even alumni can help!

        • When one of my friends was preparing her defense for a Biology PhD (something I have virtually no background in), she did a dry-run for me to see how approachable it was to a layperson. I could actually follow along; no easy feat!

        • Even in industry, I aim to prepare specific questions before the meeting, showing what I did and where I’m stuck, then if someone can’t help I can ask for a recommendation to someone that can try to help instead.

  • Don’t forget to thank everyone and follow-up, typically via email – not just the committee members, but also your supporters.

        • In industry, I aim to send an email with "action items" or add it to our meeting notes, so that everyone is on the same page on what the next steps are – and who’s responsible for what.

Dissertation

  • Scope it down! Our PhD is unchartered territory, where the goal is to propose something new. We could spend 20+ years doing that. What’s the smallest thing you can do now, to demonstrate that? (And leave everything else as “Further Research”.) It took me a while to realize I can -- and should -- scope things down, especially as a graduate student.

        • I remember I was in my 2nd year of my dissertation, applying for a dissertation year fellowship and saying that I’ll:

            1. Learn C++ to

            2. Do multiprocessing on on-prem clusters (since this was before cloud computing, Spark and the age of “Big Data”), then

            3. Develop a novel ML algorithm, and

            4. Code up the algorithm in C++, to

            5. Apply it in multithreaded fashion on the Hoffman2 cluster to analyze almost a TB of a year’s worth of Earth satellite (dense) data...

As you can imagine, that's a lot to do! My pivot (later in the year) was to scope things down by:

  1. Focusing on Jupiter’s storms captured at 1 point in time, and

  2. Develop a novel ML algorithm, and

  3. Code up the algorithm using a programming language I'm already comfortable with (R), to

  4. Apply it to run serially on the Hoffman2 cluster to analyze a much smaller dataset.

Don't get me wrong, the code still took about 1 week to converge, but if I haven't reduced scope, I may not have finished otherwise.

Defense

  • In your presentation, share why your research question is important; tell the story of how prior research falls short; and how your approach fills in the gap.

  • Did you rehearse your talk, to make sure you're within the time allotted -- and have left room for questions?

        • In industry, one of my pet-peeves is attending someone's lightening talk (~10 minutes in length) that's going at 2x because they have 30+ slides to cover.

  • End your dissertation presentation on a high note! You found something that works – even if it doesn’t work exactly as you previously expected!

        • In industry, don’t forget your "actionable findings/takeaways" (as mentioned above). Even if you found that the experiment failed, what did you learn that will help you on the next one? Mention this in your presentation!

  • You invented something new and proved it out. Don’t forget to check that you can reproduce your results! :)

        • I… didn’t know about GitHub when I was working on my dissertation from 2009 to mid-2012. Thankfully I was able to reproduce my findings after pleasantly surprising myself that the algorithm I developed actually uncovered something in the data that previous methods haven’t before. (e.g. When I applied my new ML method to Jupyter's images, it picked up on Rayleigh scattering – which helped me validate that the method was working, before analyzing the rest of the results.)

  • Don’t forget to share your defense deck ahead of time with the committee members!

        • I say this as I… sent the final version the day before – and the final, final version the day of...

        • In industry, I try to include an agenda and/or include attachments of what we’ll be reviewing when the meeting is scheduled, or ASAP before the meeting so that everyone has time to review.

Committee

  • PhD advisors don’t get paid (usually anything at all) to advise students; they do it because they want to. They’re also really busy – and don’t have time to read + give feedback on the 100+ page thesis.

        • If you give updates more often (see point above in "Communication"), and at a high-level, the updates will be more manageable to follow. I also learned to come with a specific ask/struggle/request for each meeting, to make it easier to focus meeting time and get support. This is also true for industry!

  • Don’t forget that you get the final say on who’s on your committee! And you can change who’s on it!

        • Just like in an industry interview, even informal interviews are a two-way street.

Scheduling

  • Scheduling your defense with 4+ extremely busy committee members (who are juggling research, advising many students and teaching, along with their personal life), without knowing their future commitments and availability is really hard; Doodle (or similar) is your friend.

        • When I was waiting for the last committee member to confirm the date and time of the defense for everyone, I sent out emails with only the subject line "Does X date and Y time work for you, for my defense?"

        • When I need to coordinate meetings with multiple executive stakeholders now and can’t see their calendar availability, I still follow this similar practice of sending out a Doodle or asking for a Calendly.

  • If defending in-person, confirm with your department (or similar) about availability of the room and supplies (including, projector, cables, and anything else) needed to showcase your work the day of. And find out who will be there or on-call (or on stand-by) in case of technical/logistical issues.

        • I naively scheduled my defense during IT’s lunch – and ended up running around looking for a projector at the last-minute. Thankfully a research lab not affiliated with the Department – and who have never met me before that day – was open and had a projector with cables I could borrow for an hour, to use for my defense.

        • In industry, you might find me sporting extension cables to client on-sites. :)

Parting Advice

  • Don’t forget to confirm your eligibility and submit any required paperwork to participate in the doctoral hooding ceremony!

      • I missed mine because I didn’t know if I’d actually finish that, or if I’ll end up needing Fall to prove out that my novel method works – and actually finish writing the dissertation.

  • It’s hard to know when you’ll finish because you can’t time-box inventing something new and proving it out. I wish!

  • You did it! Now, don’t forget to file your dissertation – and maybe request a print copy (!) :)

  • I can’t tell you just how much the PhD Movies I and II resonate :)

I’m rooting for you! Good luck!

You may also like:

References: