Thread by @merrierm

Some tips for grad students seeking computing research internships... 🧵

Have a professional webpage. This might be hosted at your university, on github, etc. This is different than LinkedIn (which you should also have). Link to all your publications, videos, and other research artifacts. Link to your must current CV.

Apply to the official application. After you've officially applied, it is OK for you or your advisor to send a short email to a couple of specific researchers you'd most like to work with. Briefly explain your interests. This email should link to your webpage and attach your CV.

Realize that you may not receive a response to these emails, as many researchers receive hundreds of these. Sending one follow-up is OK, but more than one gets spammy.

Avoid individually emailing every single researcher in a lab or group to express interest in them; this is also spammy and people talk to each other and realize the candidate does not have a specific interest in any one researcher.

If you have a deadline (such as an existing offer from a competing internship), share that information. Companies can often speed up their interview and decision processes if they know there is a need.

If you have flexibility (such as in doing an off-season rather than summer internship), mention this; sometimes off-cycle positions are easier to get (fewer candidates competing)

If you've met the researcher before, remind them in the email how they know you ("we met at the captioning workshop at CVPR last year...," "I attended your seminar at Berkeley in February," etc.). If you've never met, consider having your advisor send the email instead.

Researchers are more likely to notice & respond to a mail from someone they know or someone more senior (e.g., the advisor). A great advisor email will be a mini-recommendation - "My student Merrie would be an excellent intern, she is the best coder I've mentored in 10 years."

If you have an unusual skill that makes you well-suited for the internship, mention it. (e.g., you already have experience working with the companies hardware or their software APIs; you have published on the same topic the researcher is working on, etc.)

If you're offered the internship, ask questions! If you have a particular goal, make sure this internship can meet it. Do you hope to learn a particular skill? Write a research article? Meet certain people in the company? Ship a feature for a product? Make this goal clear.

If you don't get the internship, apply again next year! Many researchers prefer to work with more senior grad students who've had a year or two to learn the research ropes at their university.

Also... if you get an internship interview, prepare! Do not wing the interview! Ask the recruiter or host what to expect. Ask your advisor, former interns, other students. Do practice interviews with them. Prepare an elevator pitch about your work.

"Fit calls" are also de facto interviews. Prepare ahead of these calls with talking points and questions. Send a succinct follow-up email with links to any papers, videos, code, etc. you mentioned during the call.

If you don't know something or forgot something, that is OK to admit! (I always confuse Type I and Type II error and have to look it up. That's OK! The internet and books are useful tools! You can talk about how you'd approach a problem you can't fully solve on your own.)

I do think it's a good experience for PhD students in computing to do at least one industry internship (even if industry is not their ultimate career goal) - you learn a lot about how the tech industry works and you broaden your network of mentors beyond the university

P.S. super embarrassed that I made a few typos in this thread! If only Twitter would let me edit to correct errors... (the inability to edit for errors on Twitter continues to be a major #HCI fail IMHO)