Long but fair. There were 5 rounds. One debugging (find how to fix a bug in an existing codebase), one architecture (outline class design for a problem), one mentoring (explain a topic to someone). One with head of engineering and one with manager. Very welcoming. Quick to respond afterwards with decision.

I received a phone screen and then a take home problem. Everything is standard up to this point. Chat with recruiter. Chat with someone on the engineering team. I've interviewed at many startups.. And have worked at startups as well as big companies and this was probably my least favorite interview process. Here are the issues that they should correct for future candidates: 1) The Take home problem takes at least a full day of work to complete. As others have mentioned, it takes at least 8 hours for the simplest solution. More on that later. The problem with long exercises is that it shows a lack of awareness for a candidate's time commitments. In my situation, I had 3 other interviews that same week with 3 other coding challenges (Which were much shorter, 2-3 hours each) in addition to a full time job. Maybe this is to weed out the truly desperate or only hire recent grads. Not sure. 2) Vague problem statement. I won't post the problem here but as others have mentioned, it requires the candidate to research Markov Chains in order to solve a domain specific problem. It was actually quite interesting, but what is the company trying to gauge? Is this a problem that Brandless encounters often? Unless the developer is familiar with Markov Chains, this problem requires research into something new and research into a domain specific use of Markov chains before any actual code is written. Then layer on top of the solution the right architecture that the interviewer is looking for in a scalable web app that can be hosted somewhere. 3) The problem has dozens of solutions online. It's hard not to google the terms within the problem statement and see a solution (Even from current Brandless employees) as a candidate is researching Markov chains. I tried to be as unique as possible and test myself by not looking at any other solutions and only use the Wikipedia page as reference. In this situation, why bother giving this problem out anymore, be unique and use a smaller toy problem to gauge a developer's abilities. Although it's not a nice feeling to be rejected and it happens to the best of us, but I think Brandless has to think about what type of person they are looking for. Maybe it's worked for them in the past and they continue to do it. But evidence in the reviews of the company and the interviews here show that they need to rethink a few things. Best of luck Brandless! I think I dodged a bullet anyways.