3 ms·

Machine learning engineers will not exist in 10 years

• rvz 6y ago
  Agreed to some extent. Its quite similar to how in the 1990s that everyone was making their apps multi-threaded by hiring developers who specifically can implement a multi-threaded system when no available libraries or APIs could do this. Fast-forward today it is built into the OS, programming languages and an abundance of libraries can deal with multi-threading in a cross-platform way. Even better still some APIs are documented if they're thread-safe or not, eliminating you from rolling your own and shooting yourself in the foot. In 2010, interest in Machine Learning was a boom for researchers and statisticians; thus large tech companies snapped them up pretty quickly to build prediction models for their products. For the rest of us it was "hard". Fast forward today, libraries like Tensorflow, Pytorch opened this up to everyone who isn't a Machine Learning researcher. CoreML is built into Apple platforms as well as Google's Machine Learning libraries are built into Android. The same is already happening with AR. In this case, the engineers specialising on re-implementing or using one feature like, AR, multi-threading and machine learning will be slowly replaced by open-source libraries and will be expected knowledge by general software engineers. The 'researchers' will still exist in advancing their chosen field.