I've reviewed about 14 statements of purpose this year from students and friends applying to #gradschool across MS CS, Data Science, Ph.D. CS, EE, and AI/ML-focused programs. Here's a 🧵of what I learned.
#AcademicChatter #SOP #phdlife
#AcademicChatter #SOP #phdlife
1. You need 2-3 weeks of writing with a solid 1.5-week gap in the middle to get that 'fresh' perspective on your own professional journey. Start early!
2. Being objective* about your experience is far, far better than 'weaving stories'.
*a 3-liner with motivation, technical contribution, quantifiable outcome.
*a 3-liner with motivation, technical contribution, quantifiable outcome.
3. While a story is unimportant, a theme is DEFINITELY important. What's the difference?
Story: 'I like to peel back layers of technology and look under the hood due to an innate curiosity to understand how things work. I joined company X and learned about ResNets which I...'
Story: 'I like to peel back layers of technology and look under the hood due to an innate curiosity to understand how things work. I joined company X and learned about ResNets which I...'
...modified and applied to the problem of Y. After data preprocessing I worked on model finetuning...'
Theme: 'My curiosity to understand classifiers led to a series of experiments with modified ResNets and transfer learning, applied to a meaningful problem like X at company Y'
Theme: 'My curiosity to understand classifiers led to a series of experiments with modified ResNets and transfer learning, applied to a meaningful problem like X at company Y'
Stories are super personal, themes aren't necessarily so. Being personal has its advantages, and it's a 'right-place right-time' thing. Hard to generalize where it will help, but if you've dealt with hardships, a sentence is all I would add, given you understand what hardship is!
Themes, however, are almost always super useful eg evaluating if your work aligns with those whom you want to work with.
Note: One theme is all you need; even if you have five potential research directions, pick the strongest one and build an excellent case for it.
Note: One theme is all you need; even if you have five potential research directions, pick the strongest one and build an excellent case for it.
How to spot a strong theme?
A theme should clearly indicate the 'purpose' of your statement. A purpose CANNOT be just 'to gain knowledge' bc that reveals nothing about you compared to a Tom, Dick, or Harry. What is *your* purpose? Why is it unique*?
*
A theme should clearly indicate the 'purpose' of your statement. A purpose CANNOT be just 'to gain knowledge' bc that reveals nothing about you compared to a Tom, Dick, or Harry. What is *your* purpose? Why is it unique*?
*
https://twitter.com/swapneel_mehta/status/1446889288868712454
4. Read what professors are looking for. It's a new age, and papers are backed by blog posts, YouTube videos, Zoom presentation recordings, and tweet summaries! Have an idea of whom you want to work with and why, before dumping names into your statement.
5. Read every single sentence you have written and evaluate if that reflects uniqueness. If not, either drop it or rewrite it. See a thread on why this is important (tl;dr, you are selling yourself, this is a sales tactic)
https://twitter.com/swapneel_mehta/status/1446889270829060097
6. Be genuine. It is not easy to magically come up with a meaningful life goal and that's ok. But it is not ok to not even try. Think about topics X you are best suited to pursue, and topics Y that you really *want* to pursue. Then, find overlaps or at least connections.
Everything you say you *want* to do or *will* do WITHOUT historical evidence in pursuing related tasks makes your application look incomplete. Just focus on what you have done 85% of the time and what your plans are BASED on this work, 15% (arbitrary split, but you get the idea)
The statement of purpose is the last thing that is usually used to evaluate you after passing some threshold for GPA, GRE Scores (may not be required), etc. and you will be competing with really strong applicants. Keep that in mind so that you use the 2-page real estate wisely!
7. You are not perfect. Nobody is. And it's ok. But talking about negative things is unnecessary. Focus on the upside. There is almost never a good reason to show how things went wrong for you and you managed to make it. Often, such stories either sound mundane or contrived...
...despite your best intentions and do not work to your advantage. It is ok to mention technical challenges, just don't beleaguer the point. Focus on quantifying your impact and highlighting your learnings!
To quote my wisecracking self from two years ago,
"There are no bad experiences. Only great experiences and great *learning* experiences!"
"There are no bad experiences. Only great experiences and great *learning* experiences!"
8. Lastly, try to avoid throwing in random famous quotes unless absolutely necessary. Ok cool, you read, follow podcasts, and watch YouTube videos, but unless this changed your life and put you on a meaningful career path, it tells me nothing about you really!
Instead of just people, focus on (their) work that may have inspired you. Don't just cite work, offer a one-line summary and suggest what directions it relates to. Maybe it helped direct you into professional opportunities in the area. To me, that's more reasonable than quotes.
This is just a partial set of opinions and I have many more but I am also tired and hungry and need to get back to work. Write well. Goodbye.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
