Summer Work App: Fill & Download for Free

A2A. Speaking only about Top-100 or so CS programs in the US. There are a few scenarios in play here, as there’s still a lot of decisionmaking to be done once you’re in a Ph.D. program that determines how much you make. We’ll start with a baseline minimum, and it really depends what you add. You’ll see that there can be a lot of variation, depending on how hard you’re willing to work (and how lucky you get).Minimum: No fellowships/scholarships. Work only during the school year (~18k-30k). This will probably be called a stipend or assistantship (research/teaching/etc).Things you can add:summer work in the lab (add 6k-10k)summer work as a public sector researcher (add 0k-20k/ yr)summer work in the private sector (add 10k-40k/ yr … I think 40k is unusual, but I have a colleague who pulled it off at a startup)publish your app and collect on it all year (add .5k-100k/ yr)win a fellowship (add 2k-16k/ yr). Note: winning a fellowship usually doesn’t add atop your assistantship, but will only supplement it a small portion of the total award amount (the school will take some).pick up a tutoring gig atop your assistantships (add 1–5k/ yr)… I think I’ve hit the most common ones. As is probably obvious, you can’t do all of the things on this list. For example, you could easily lose money on a published app, you may win a fellowship that disqualifies you from summer work, or your advisor might “highly recommend” you try an internship, which means you won’t be doing lab work that summer.

You could try your hand at autocomplete! This is a fun problem that is very flexible--you can bang out a decent solution in a weekend, but if you decide to keep going with it you can continue making progress for another week or two, or three... It is also a problem that comes up all the time in web applications.Say you have a list of keywords that a user might search for. Each time the user enters a keystroke into the search bar, you want to return a list of all the keywords that begin with the search query. For example, if your keywords are 'apple', 'donkey', 'aardvark', and 'apparent', then you want to return the list 'apple', 'apparent' when the user has typed in 'ap.'The most obvious thing to do is just maintain a list of all the keywords, and scan through the list after every keystroke to pick out the matching keywords. But what if there are a ton of keywords--say 50,000 or 250,000 or a million? (For example, the keywords could be the full names of all the users of an app.) Then the list-scanning turns out to be too slow, so you'll need to be a little more crafty with your data structure.What if you want to implement fuzziness? So that if a user has a typo in their search, like 'prgrammi', you still return the keyword 'programming'.Once you get the speed sorted out, you can start to worry about the space-efficiency of your data structure. Objects take up a ton of space in Python; even an empty dictionary occupies 280 bytes. Depending on how your structure is implemented, there may be lots of room to optimize its space efficiency by reducing the object count.Lastly, you can start thinking about how you would incorporate your data structure into a web app. Would you rebuild it every time a user visits the search page? Would you pass the structure to the browser on page load, or would you ping the server after every keystroke? Would you cache it? Would you compress it into a blob? You might compare the performance of the pickle, cPickle, and json libraries in Python.In short, I highly recommend autocomplete as a fun and challenging side project! I spent a decent chunk of my summer working on this problem, and it was definitely time well spent. You don't need much info to get started, but there is a ton of room to improve, and you'll learn a lot along the way.

Reggie Brown, one of Snapchat’s founders, had sent a nude photo to a girl, and he was regretting it. He said, “I wish these photos I am sending this girl would disappear,” . Afterwards when they were talking, Spiegel had heard about the idea was very excited about it. He said it was a million dollar idea, turns out it was quite the understatement. Anyway, they launched the app and called it Pictaboo.They spent the summer working on it, so it eventually became the Snapchat we know and use.