HN user
This is great!
We run Quizlet behind Cloudflare, and use Google Cloud for all our server infrastructure (>150 VMs). We've been very happy on both platforms. We'll be saving around $2k/mo on bandwidth because of this deal, and we didn't have to lift a finger. Yay :)
Happy to answer any questions about either platform.
Quizlet is hiring in SF! We are looking for designers, software engineers, infrastructure engineers.
Quizlet is one of the rare companies to have achieved scale in K-12 education. Our team of 11 engineers ships code to 22 million students and teachers per month. We are a bootstrapped, mission-driven company with a goal of building amazing studying and learning tools to the world. We spend a ton of time talking to teachers and students, through our feedback center [1] and by visiting local schools. And we've barely gotten started on all the great tools we're planning to make.
Contact: jobs@quizlet.com
[1] http://quizlet.com/inside-quizlet/quizlets-incredible-feedba...
Thanks a lot! The polish came from a hack night we did to redesign it.
I don't think there's much to open source -- it's about 2,000 lines of code. It's mostly just writing custom sql and elasticsearch queries to show data related to the users we're about to respond to.
Re: the log integration, all our application logs get written to logstash/elasticsearch/kibana, and one of the fields on the logs is the person_id (they're unique id whether logged in or out). When we show the logs in the feedback center, we just query elastic search and group by the different log types.
Quizlet (http://quizlet.com/jobs/ios-engineer) - San Francisco, CA - iOS Engineer
We're looking for someone to start and lead our iOS team as our first iOS engineer. We've put together a killer team across web, infrastructure, android, design, support, but iOS is still open. We would love someone strong enough to lead a team that should grow to 3-4 in the next 6-9 months.
Quizlet helps students practice what they're learning. Students absolutely adore us, just read the twitter search for Quizlet. Quizlet is a top 100 Website in the U.S. (20m uniques/mo), and during school our current iOS app is always in the top 2-3 for education. We are one of the few education companies succeeded in going directly to students, as opposed to selling through schools and districts. We have big plans for building learning tools that truly empower kids and make them enjoy the learning process.
Applications to jobs@quizlet.com
no, maybe we'll try it.
Thanks for the note -- the performance problem was in set(), not get (that was lazy of me).
We didn't have a lot of triggers running -- we spent a lot of time looking at performance traces. Just initializing models on page load was incredibly slow. It was particularly bad on a retina iPad.
No we didn't.
For our recently launched flashcards tool, we built it in Backbone. Demo: http://quizlet.com/23770911/flashcards
Our experience was that backbone was very helpful in laying out code and providing clean event triggering, but that a lot of it was simply too slow to use. We started out with a View per card, but that was impossibly slow with 100+ cards (and we need to support ~1000+). We ended up with just one view for the Card section, and a view each for the sidebar and options sections. That made it much faster, but also took away many of the benefits of backbone.
We also ran into significant speed problems with model gets and sets. Initializing the data for 200 cards (~10 attributes per card) took hundreds of milliseconds on Chrome and other browsers. We ended up using plain javascript objects for the hottest parts of that code.
Overall to us, backbone was probably worth it, but also cost us many long days of performance debugging, and we're still not 100% happy with it. For our next project we may try something else.
Quizlet - San Francisco, CA.
Lead iOS engineer - we're looking for a person to lead the development of our iPhone and iPad apps.
We care deeply about education and do a lot of unique things to get in the minds of our users (16M uniques/mo, mostly high school and college kids).
- Every week we attend a different high school or middle school in the San Francisco area, where we test new features or ask for improvements. Everyone on the team attends.
- We personally respond to 1,200 feedback messages a week. Everyone on the team pitches in. It sounds hard but our tools are amazing for this
- We use our product a lot. Our team of ~12 combined knows about 10 languages, and groups of us are always taking classes or trying online learning tools. We take group tests (last week we all learned Sushi terms).
Ping me if interested: andrew at quizlet
I was at a big edtech conference last week (Education Innovation Summit in AZ).
Something like a dozen of the 100 presenting companies came through Imagine K12, which is incredible. I know a lot of the people involved, and if you're thinking of starting something in edtech you should DEFINITELY apply. It's a great program.
I believe it's the Mexican flag
The logging thing is really really dumb. Having logs in a real database lets you do incredible things (e.g. correlating user feedbacks to random problems you noticed). We use our db-driven logs many times a day to debug issues and find patterns.
Don't be afraid of putting logs in databases and then figuring out how to scale it. It's not that hard and it's very worth it.
Quizlet, SF [Full-time, Interns]
This is a picture of two kids using our unreleased learning game: http://qdaq.com/4j1.jpg
We took this photo in a classroom last week where we were beta-testing our educational game. We want to hire engineers who are excited by helping kids learn, and creating similar reactions to the one above on a scale of millions of kids. We're building web and mobile software that makes a significant difference in the lives of its users.
More info: http://quizlet.com/jobs/
Ya, logs are soooo much more valuable when they're in a database that you can query. We log to MySQL and then correlate logs with users, events, IPs, pages, etc.
Logs are just as relational as anything else, it took me awhile to realize it though.
Ya, I'm saying if I had done any googling, I might have found flashcardexchange.com or other competitors that existed when I started it, and settled for one of them. I'd say it was luck that I didn't do any research, and when I decided to build my own thing, I had a clear vision of what I wanted.
All this was when I was 15, and I wasn't thinking about things in terms of startups and competition and so on. It's certainly possible that another way it would have happened was researching the competitive landscape and deciding I could do it better.
I don't see our competitive position as different than any other typical company. We've got a great product that people love, and we build new things into it that our competitors can't predict or move fast enough to build. We've built a brand that students and teachers know and like.
One deliberate competitive advantage is that we're a free, ad-supported site. Most education products go for a subscription model, and that requires enterprise sales (selling to districts, blech) and fundamentally limits the number of people who will use and enjoy your product.
The ad model only works when you have scale (which we do). If we had been in a position where we needed to make money quickly, we wouldn't have grown into the size and scope we have now.
Quizlet was originally for myself and then for my friends in school. It wasn't even a real business for over a year. I never did any market research or surveys. I just built features I wanted and features my friends wanted, and it turned out to be a thing that a whole lot of other people happened to want. Quizlet has grown 100% through word-of-mouth since it started.
Not saying those things aren't useful, but nothing beats being your own customer. Since I'm not a student anymore and still work on Quizlet full-time, it's super important for me and the whole team to have exposure to our users' experience and mindset. So to hack it, we started a Spanish class at the office so we'd have to use the product. We've also been going on field-trips to various middle- and high-school classrooms around the bay area about once a week.
FWIW, if I had done this before I launched Quizlet, Quizlet wouldn't exist. In fact, if I had even done a google search for the functionality of Quizlet, it wouldn't exist. Now it's 5x bigger than any of its competitors and a top 400 US website.
Just sayin.. :)
San Francisco, CA
Quizlet is a learning platform for students that puts students in control of their own learning. We're used by millions of high school and college students and are a top-400 website in the US.
Quizlet wants to build a next generation of learning tools that enable students to deeply and creatively learn whatever they need and allowing students to be the source of their own curriculum.
We're looking for developers interested in working on the entire stack, but specialization is ok too. Check out our recent Nodejam win for a look at what we do: http://quizlet.com/blog/an-epic-how-quizlet-won-nodejam/
From the time I arrived at MIT in '08 to the time I left in '11, I witnessed a pretty huge shift in the interests of MIT students towards startups. Many people who probably would have gone into finance or defense diverted themselves to the startup world. The 100K competition at MIT and things like StartLabs have gotten big. Dropbox being homegrown was a big thing.
For me, it's been awesome to see the best engineers doing their own companies instead of joining defense contractors or GS.
Just doing DNS checks might not be enough:
`$ dig hotnail.com MX` yields results but is still invalid.
ya, this. There's literally nothing jQuery related in this code. The jQuery part prevents it from being run in node or in non-jquery sites.
It's a cool idea, but I think they're doing it wrong.
We experimented with doing something like this on Quizlet, but didn't actually launch anything. We first looked at a lot of the data and doing based on string distance is the wrong approach.
For example, if you type hotmail.de into that checker, it suggests hotmail.fr. Another is ymail.com --> gmail.com. The more valid domains you add, the more (correct) permutations get marked as invalid. We have 20k users with ymail accounts.
I think a blacklist approach is much more solid than a whitelist approach, I just haven't gotten around to building it.
All of this is easy to auto-detect in Javascript that you can have on your support submit form--so this page is not that useful.
What is KILLER useful is http://showmewhatswrong.com. Instant screencasts from users of what is going wrong.
this can be greatly improved (losslessly) using http://punypng.com / http://www.smushit.com/ysmush.it/ / downloadable PNG recompressing software. I tried it on several files on this site and some decreased in size by more than half.
More examples:
Swedish numbers: http://quizlet.com/3614707/speller/
German clothing: http://quizlet.com/1692182/speller/
French animals: http://quizlet.com/1060977/speller/
Japanese adjectives (good luck!): http://quizlet.com/343188/speller/
We decided the most realistic audio came from commercial vendors.
In terms of mobile stuff, we've thought about doing our own but for now there are great apps being built on our API. We just launched V2 of our API with OAuth integration: http://quizlet.com/blog/quizlet-announces-api-2/
If anyone's interested in building mobile stuff, feel free to play around with it. Or contact us (email in profile) and we may have a job for you :)
Rasmus gave a talk yesterday and I asked if they were switching to github. He said they'd likely switch to git (it's now happening, apparently), but not github because it can't do ACLs and the PHP project has thousands of people who need commit access to various places.
Yea -- when we have arabic text on Quizlet, we override the font to this:
font: 2em 'Al Bayan', Nadeem, 'Simplified Arabic', 'Arial Unicode MS', sans-serif;
When I took a speech rec class last semester, we had a guy from BBN (subsidiary of Raytheon) give a talk about large-scale, extremely fast audio transcription. As in, systems that could process audio 30-40,000 times faster than real-time. They traded off recognition accuracy to get this, so their accuracy was around 50-60%, as I recall. I asked why something like that would be useful, and he said if you're looking at a lot of data (which I heard: eavesdropping on an entire telephone network) then all you need is a general idea of what people are saying and a few keywords before you can zoom in on specific clips for more thorough analysis.
So yes, I'm sure NSA is interested.