Note
This is a permanent ad, up even if we do not have any budget. We do this for two reasons:
- if management gives us budget, and we have you as a "yes" in our database, we can make you an offer, settle your visa and book your flights (which takes us around a week) before they can change their mind;
- if you are exceptional, we can pitch you as "we have to take this guy off the market, even though we don't have budget right now". This has worked twice already.
Summary
We are an online retail shop with millions of customers and 9 digit USD funding. The data science team is a group of 5 Haskellers who control the (almost purely functional, Redshift/Nix-based) data mart and help other departments solve hard problems. It's a crossover between the traditional insights based work, and software engineering, as often a durable solution will need production code. We like to hire smart people and worry about the job description later.
Details
Location: Singapore. No project management structure - work directly with the users. No corporate bureaucracy - wear shorts to work and turn up at lunchtime if you want. Remote work might be an option, help us pitch it to our traditional minded management by telling us the stuff they care about like years of experience and brand names.
This being said, we do not care about your CV, years of experience, college education, cover letter or hobbies. We estimate ability, and will measure your future performance, based on your code. As such, please solve the task below in your own time and email your code and results to corentin dot roux at zalora dot sg. I usually respond within a day.
Task
If you are wondering why we put up a task, read "Delta Force" by Charlie Beckwith (management perspective) and "Inside Delta Force" by Eric Haney (data scientist perspective).
Here are 453 papers (as of this writing) containing recent IT bubble words like "big data": http://tinyurl.com/mczfmz9
Here is a paper: http://arxiv.org/abs/1309.4168
Crawl the first link to download all the papers, make them into a corpus, apply the methods in the second link, use PCA to reduce it to two dimensions, plot the words in question, send us the code for everything, the corpus, and include the graph in the body of the email. Try to write as much as possible in Haskell.
