think-cell C++ Developer Interview Questions
& Process

I took an assessment in April but didn't pass. I improved my skills and retook it, passing on the second try. However, they said my application wouldn't move forward because of my previous application. My first application was for an internship, but this one was for a full-time job. It's a bummer that an old, failed internship application would affect my chances for a full-time role, especially since there was a big time gap. I reapplied because I really liked the programming test and wanted to see my improvement since last time. It's frustrating that my rejected internship application was the reason I didn't get the full-time position.

I applied online for a remote C++ developer position. The next day, I received a link to a coding test that required implementing a single member function of a class within a 9-hour timeframe, though 3-4 hours should be sufficient for those familiar with the concepts. The challenge involved specific, less common restrictions related to the 'gang of four' design patterns. You could submit your code twice. The first submission provided general hints, and the second offered details on which test category failed, along with examples. I missed one restriction, which was shown in the example. Familiarity with these topics would likely reduce the time significantly. The coding environment suggested using an IDE with g++ 7.20 on C++17. I used VSCode with a newer g++ in WSL and verified my code on Godbolt to ensure it worked with the specified compiler. Writing thorough tests and preparing the environment are crucial. Although I didn't pass, the task is doable, especially considering the salary range, and not as trivial as stated.

I failed at the first stage, like most people. I thought implementing a data structure wouldn't be hard, but after wasting a whole day, I wish I had listened to the reviews. I wrote a wrapper class using a known library to handle the problem and compared results. This wrapper tested invariants, in-range/out-of-range values, and more using random numbers. I ran at least 1000 random numbers per operation, with 10+ operations per test case, totaling 12 test cases. I even stress-tested with over 1 million operations. Despite this thorough testing, I couldn't find the specific corner case they were looking for. It seems they want the top 1% of C++ programmers. My advice is, if you're not an experienced C++ dev, save your time. For experienced devs, the challenge is finding that niche corner case. It was a tough but solvable problem, and I didn't succeed.