A good Statement of Purpose (SoP) answers two key questions:
When you clearly address these points, you naturally answer the most important question: Why you?
In general:
Start as early as possible on the first draft.
In my opinion, an SoP should serve (strictly) as a research statement. It should focus on describing your past experiences and explaining how they connect to what you want to do in the future.
cs-sop contains many great examples that helped me a lot in drafting my SoP. I highly recommend it. It offers an excellent overview of the general framework, structure, flow, and expected content.
From my own experience proofreading students’ and friends’ statements, here are my two cents:
By the first few sentences—or at least by the end of the first paragraph—readers should understand your overarching research goal.
Keep this research goal as the main theme throughout your SoP.
If you have a concrete research goal, things become simpler. The rest of your paragraphs should explain why you can accomplish this goal. In other words, discuss how your past experiences have prepared you for it.
Highlight the following:
Try not to repeat information already listed on your CV. Most universities have some page/word limits, so use your word count wisely. Avoid making your SoP a paragraph-long version of your CV.
If you look at the SoPs on cs-sop, you’ll see that the structure / content of this section is fairly similar across statements.
So what make your statement stand out is to customize it as much as possible for the particular lab or PI you’re interested in. Make sure you understand the lab’s most recent research directions, which might differ from their earlier or more well-known work. This also helps you find out if the lab is a good research fit for your goals.
Cheers.
Update Sept 2025: New insights and comments are added to my original 2024 version with colored text.
I’ve met a lot of people who have done or are doing a PhD for all sorts of reasons - some wanted to become professors, others aimed to join a top industry research lab or build a startup around their research. I’ve also met people who pursued a PhD without a “dream career” in mind, but who had still carefully weighed the risks and rewards. Whatever your reason, it’s valid as long as you understand your goals and are ready to commit.
BTW if you choose to do PhD NOT out of passion (which I’ve seen a non-trivial number of such case), you should think twice. First, you should still enjoy the day to day work if not love research to not suffer. Successful PhD often come with passionate people who also work hard. So the mindset of “I’ll be able to grind through this” is not efficient and sustainable in the long run. It is essentially an optimization problem :) You should find a shortest(er)-path towards your goal :) There can be (and often are) many ways to reach the same destination, and a PhD might not be the optimal one.
I thought about these three questions when making my decision:
Upper Bound: Can I do what I want to do without a PhD? Will five years in a PhD help me reach goals I couldn’t otherwise achieve in the same timeframe?
For instance, if you want to be a professor, the answer is usually yes—you need a PhD. But if you’re aiming for a research role in industry, there might be other ways to get there without a PhD. Eg: I’ve met folks who are able to transfer into research PA (org) as a research engineer at Google after a few years of experience.
Lower Bound: What if things don’t go as planned during the PhD?
Are you prepared to handle the worst-case scenario, like switching advisors, switching research direction, or even leaving the program? For some people, losing a few years of industry-level income and work experience can be a significant downside. Make sure you’re aware of the opportunity costs and sunk costs.
“Expected Value”: can I live through the day to day as a PhD student?
PhD student can be challenging because of the opportunity costs (trade-offs) and the lifestyle. You’ll face rejection, stress, confusion, and slow progress on research. On the flip side, you gain the freedom to explore new ideas, develop deep expertise, and learn resilience. “a PhD degree shows that you can handle tough situtation” by Ray from my conversation with him when I was deciding if I want to pursue a PhD.
One thing that helped me in determining if PhD is right for me is to talk to A LOT of people who has went through this journey, or part of the journey. Roughly 50 1-1s with people during summer of 2023.
Ultimately, I decided I wanted a PhD because:
I might not be the best person to answer this question. I decided to apply for a PhD at the end of summer 2023 and applied in the same fall. I didn’t have much time to prepare specifically for the application process. Thankfully, I already had some research experiences and received a lot of help from my Master’s advisor, Akshitha, and other friends in the lab (Jaylen, Edwin, Veronica, Sara), which I will be forever grateful for.
Being a direct beneficiary of mentorship, I hope to pass down the help I have gotten to people out there.
Doing a research internship is a great way to: 1. get research experience / LoC, but 2. more importantly, the first hand experience of doing research so one can know if research is a good fit for themselves.
If you are an undergrad without research experience, don’t hesitate to reach out to PhD students and faculties. The worse case is that you don’t get a reply or a rejection, but you’ll be “rejected” if you don’t even reach out from the first place. I was a very shy person so I understand that it takes a lot of courage to even take the first step.
One thing that I did not know before PhD application is that many labs takes undergraduate research assistant both fulltime and parttime. It is a thing.
You don’t need to attend the same university where you do research. Many labs welcome undergrads from other schools if you have the right skills or interests.
One thing I did is that I made a execel sheet listing out all the insitutes, departments and research lab / PI that I wish to apply to. Doing it takes a lot of time, but it was very helpful for me to:
Try to research online about the lab. Some red flags are easy to spot, but more requires you to talk to people. Try reaching out to PhD students at the lab you are interested in to find out more about the latest status of the lab, and whether if the lab is a great fit for you.
Know your order of importance when picking PhD program, it is human nature / undergrad mindset to pick university based on ranking or prestige. But PhD program is very different, a great advisor fit is far more important than the insitute’s ranking. So don’t limit yourself when looking for labs and PI to top ranking universities only.
Checkout my other blog on PhD SoP. But one resource that helped me greatly is: cs-sop.
The content is the most important thing. But besides that - START EARLY!, AS SOON AS POSSIBLE!
I completed my first draft of SoP by mid/end of October, and I spend the rest of the time polishing my statements. I asked a lot of people to read my statements. And I max out all the writing center appointment (one/week) that I can possibly get at CMU. Even then, I had to make last minuet changes on my statement an hour before the application deadline.
In this process, it will become more obvious how to connect the experiences you have, and what you want to do going forward. The processing of telling your story to other is a great way for yourself to organize your thoughts.
Btw, this is a good opportunity to think about: 1) whether the work you want to do fits within a PhD program, and 2) why you want a PhD. For example, if you are interested in working on pre-training LLMs, the resources for doing that in academia are definitely more limited than in industry frontier labs (as of 2025). I’ve been reached out by people asking for help framing why they want to do a PhD in their SoP, but in my honest opinion, you shouldn’t pursue a PhD if you don’t know that answer for yourself.
In preparation of the interviews, I created a slide that contains the following content:
This will be helpful both for faculty who interview you, and yourself during the interview. Be ready to conduct the interview with and without the slides. And you can ask the interviewers if they prefer slides or not at the beginning of the interview.
Interview Questions:
Different advisors ask for different questions, but the typical ones are:
you are about to make a long term committement, you are about to work with these people in the next five years. It is your one chance to find out if the lab is a great fit for you or not before accepting the offer.
During my undergrad, I had minimum to no interaction with people who are doing PhD, considering doing a PhD, or had a PhD (recent PhD grad, PhD student, grad student, or senior undergrad, or close interaction with professor). The idea of doing a PhD also never cross my mind until I met Akshitha during my Master’s. I hope this post can also serve the outreach purpose for folks out there who doesn’t have such exposure. Discliamer: I did my undergrad at UofT, one of the top research universities. There should be (and probably were) a lot of opportunities for undergraduate research, but still, I was not exposed to research research as I was not actively looking for such opportunity, so you can imagine how hard it would be for student from under-represented university to get exposure to research as an undergrad.
I realize I’m not able to keep up with every individual request. If you have questions, please share them through this link, and I’ll do my best to answer them in batches on a regular basis.
Best.
]]>