I applied online and after 3/4 weeks I was contacted by a recruiter. The process was initial call with recruiter, call with a team lead, take home test, technical and system design interview and finally HR chat. The take home test is open ended and takes a lot of time if you want to do it properly, I spent nearly 12 hours. Starling has no respect for the time of the people that are interviewing for them. After putting so much effort and time, after my technical and system design interview I was told that the position had been filled. I found out from a friend who applied for the same position that he got the same email saying that the position had already been filled at least a week before I had my technical interview. If Starling had let me know earlier I would have not wasted hours and taken time off of work to do their pointless take home test and prepare for their technical interview. Really disappointed for how they wasted my time.

The interview have 4 stages: 1. Phone call with one of engineers - passed 2. Take home excersise - expected to be completed in 3-4 hours, but more about it later - not passed 3. Call with engineers to discuss the excersise 4. Call with senior lead Let me focus on the take home excersise. The expected time to completed the excersise is less than 4 hours. The excersise is about creating a REST end point, where you have to collect the data from the Starling public API, transform it, publish data to the Starling system and return response to the call. You are given nothing more than very vague business requirement and how to access their public API. So you need to understand requirement make yourself familiar with API, what endpoint you would need to call, build REST clients with all boilerplate, and tests (I was using WireMock). Then you need to configure your project. For example you need to configure HTTP header to send API authorisation token. But from perspective I guess I should hardcode it as using common pattern of having config file was brought in feedback as "confusing how it is done". So it took me good day of work to have all set up, before I could start focusing on actual business logic. Another few hours to actually write the business logic code. As I use TDD that also include tests suit. At this point you need to make compromises if you don;t want to spend another day to make the application robust. I make compromises, which I was fully aware on error handling (I put that even in the readme as futher improvement), documentation, not making the app really functional as that would require adding persistence layer (another few hours). I knew that the solution was not completed, but you have to stop somewhere and I would expect that in the next round someone would actually ask me about those comtpromises. But I had no chance to defend myself as I was rejected without going into next stage. I've used one of the the Java frameworks (SpringBoot, Quarkus) which abstract a lot of logic behind those steps, and which I would use in production level solution. But it happened to be a challenge for the developer who reviewed my code. So don't use those frameworks!!!! Just write the code like you did when on university!!!!