TL;DR: They want Data Engineers who have completed an online ML course. I only enjoyed speaking with one person; the rest of my interactions were the worst of what tech has to offer.

Their success comes from their sales team (which I don't begrudge – stop pretending your ML algorithms are the core of your success). They have no diversity, a culture that seems to be based on an implicit hierarchy, and they expect a techbro culture turned up to eleven.

They are probably going to face harassment cases or, at the very least, some kind of workplace bullying. They lose people to Canva constantly (though not in reverse). Instead of hiring local talent and teaching them what they need, they throw interviews that would probably work in SF/NY, failing to realize that web-scale developers are much rarer in Sydney.

Finally, I decided to write this review (the interview was like a year ago) because I ran into a former Rokt recruiter who told me I dodged a bullet and then mentioned that Rokt has a pending lawsuit over harassment.

The process went as follows:

1. They sent me an online personality quiz, which I refused to do. This should have been the first red flag.
2. A third-party recruiter reached out to me, at which point I got to speak to an Engineering Manager (who was the only person I liked in this entire process).
3. I had a choice between a take-home LeetCode or an in-person LeetCode. I chose the in-person LeetCode. They were non-trivial, especially for Sydney-based tech, which doesn't do much LeetCode. I imagine a lot of people get rolled by this round, which is why they've been recruiting ML engineers for the past year. I got all of them right with time to spare, but I was a bit lucky there. I will say writing tests for your LeetCode is an incredibly effective way to develop fast (use pytest in your IDE, not the built-in HackerRank garbage).
4. I had an in-person coding interview with the Engineering Manager, which was much simpler and more on par with what I think the level should be. We had a good discussion about coding, and I talked a bit about how I had impacted my previous organization. This was the only good experience I had, and I hope that Engineering Manager finds a better company.
5. The ML interviewer appears to have completed an online course on ML, and it felt like I was going through a checklist of ML topics. If I went a bit deeper into anything, he instantly changed the topic and didn't engage. We covered a large amount of surface but with no depth. You could honestly get a competent software engineer, sign them up for an online ML course, and they'd probably pass. By far the lowest level of ML interview I've done. I ask my interns questions that go more in-depth than what Rokt did for me.
6. Systems Design – this was a bit of a disaster. I have no web-scale background, so when I was asked to design some kind of system, I had no grounding to work with. I talked about accessing data but was then told I wasn't allowed to access this data since it belonged to another team. Apparently, the solution was to use S3 as my backend? It was a situation where you had to know precisely how to build it, or you had no way to arrive at an exact solution. Instead of getting help, I was belittled for not knowing, told my smaller-scale experience was irrelevant and basic, and that the industry I worked in was very primitive. He ended up saying that they usually only hire from AWS and Google because no other tech firms in Sydney have the right experience. I also got quizzed on basic databases after that, but again, when I asked him some deeper questions, I got no answers.
7. So, despite failing Systems Design, I was passed through downleveled (one below Senior) to the CTO interview. I was told we didn't need to waste anyone's time and I could just pull out of the interview. I was a bit vengeful and scheduled an interview at an inconvenient time for the CTO (in NY). Upon doing so, I realized the company is actually a sales firm – like 300 sales people to fewer than 100 engineers, and the ML they are proud of is pretty much a very basic thing (hence the terrible ML interview).

When I talked about wanting a larger company to grow, I was told they had a scrappy startup culture and only those with drive and motivation would succeed. I was a bit gobsmacked by this – you're a 400-headcount company; you're too large for that model to work; it should already be breaking down. And I'm wording this a lot nicer than what the CTO said. I've worked at Investment Banks and Management Consulting firms with better cultures.

Anyways, they didn't even email to reject me; my recruiter had to call them up to ask. I think I was effectively rejected with my downlevel, but I didn't take the hint.

Systems Design for how Rokt's product works.

Think how Amazon recommends products when you add something to your basket or look at a product.