50%+ of my time is spent on backend engineering because the ML is used inside a bigger API. I take responsibility for the end to end experience of said API, so

50%+ of my time is spent on backend engineering because the ML is used inside a bigger API. I take responsibility for the end to end experience of said API, so I will do whatever gives the best value per time spent. This often has nothing to do with the ML models.