Coinbase needs to fix this before their IPO. How are they still not prepared for traffic surges?
On my first interview I used C++, and it was very hard to pass their expectiations (the interviewer didn't know C++, so he couldn't appreciate the effort of me using a low level language).
I switched to Ruby for the next rounds, and got through easily.
Then I went to their beautiful office, I loved the people and the culture there, but didn't code fast enough to join the startup (which I'm happy about...startup mindset is not for me).
I don't think impressing an interviewer is a good reason to pick a language for leetcode type questions..
In another interview at a quant hedge fund I had a lot of fun talking to the interviewers (even the CEO) about low level data structures, and I got some tricky questions about the CPU that I really liked.
I don’t know if they treated you fairly or not, but the policy at least of appraising an instinctual reach for low level C++ as a negative sounds correct to me and should be a much more widely used heuristic in software hiring.
But then you started talking about static typing as negative, and suggesting to move fast and break things for a company, that probably hadles trillions in assets.
Most of my career has been focused on helping large companies use Python to achieve as-fast-as-C performance with extremely little code and extremely safety-focused non-OOP software architectures, including in high frequency trading (where I built a company’s entire core trading platform in Cython).
The idea that static typing leads to safety is false. Static typing adds more code, more code is more liability. The cost of that liability is not recouped with any offsetting benefits.
When I applied to a quantitative hedge fund, I was being asked questions about synchronizing data structures between threads without using mutexes (again not something you often do in a high level language). I was accepted there, although I couldn't go to work there because of health reasons.
At the end I went back to Google where I had to launch a new machine learning model without introducing even 0.1ms latency in the ads backend. Good luck doing that in a high level language.
Edit: lol what a day to say that! For the first time I’m seeing unplanned outages on Gemini.