I’ll always remember the primary time I received a PagerDuty alert telling me that mannequin scores weren’t being returned correctly in manufacturing.
Panic set in — I had simply accomplished a deploy, and my thoughts began racing with questions:
- Did my code trigger a bug?
- Is the error inflicting an outage downstream?
- What a part of the code might be throwing errors?
Debugging dwell methods is aggravating, and I discovered a crucial lesson: writing production-ready code is a very totally different beast from writing code that works in a Jupyter Pocket book.
In 2020, I made the leap from information analyst to machine studying engineer (MLE). Whereas I used to be already proficient in SQL and Python, working with manufacturing methods compelled me to stage up my expertise.
As an analyst, I largely cared that my code ran and produced the proper output. This mindset not translated effectively to being an MLE.
As an MLE, I rapidly realized I had to give attention to writing environment friendly, clear, and maintainable code that labored in a shared codebase.