I want to make a change next year. I have been working on Front-end with the MEAN stack (MongoDB, ExpressJS, Angular and Nodejs). But I keep seeing exciting projects with AI/ML I want to work on full time on that. I don't know where to start. Any suggestions about how to start with it and how to develop experience on that field to be "hirable"?
^ assuming you don’t have PhD, do that + at least 3-5 years of legit coding experience
Is all the book/paper learning useful (particularly deep learning)? Seems that interviews consist of system design, and questions about recommendation systems, ad targeting, and generic stats/data science. How does one get past the resume filter?
you will be grilled on which models to use when, why, what kind of feature engineering to do, which optimizations methods to use... also just to gain general fluency and to sound like you understand the “language” of people who do AI/ML the link to the reddit post is basically the bare minimum to get and pass interviews for ML positions at FANGy places
also learn backend infra. if you’re into thinking via “stacks” check out the SMACK stack
How helpful is infra experience for landing and being successful in ML Engineer positions?
it’s more helpful for being successful, though some ML positions lean systemsy there is a lot of data engineering involved, and ML models / methods are implemented almost exclusively in backend. since they can get quite expensive (both CPU and memory + bandwidth), knowing your backend and infra constraints can completely change which approach you take to tackle a problem i’ve been asked questions like which type of regularization to use if this model is going to be part of a realtime vs. batch system, etc. https://developers.google.com/machine-learning/rules-of-ml/ ^ also amazing resource and gives you a sense of why backend and infra experience is crucial. the bottleneck to applying ML in most cases is engineering, not theory, though you need an understanding of both
try some Kaggle competitions, brush up on Python, Java, C++, SQL
get a phd degree of ml
Does everybody working on ML have a PhD? I doubt it that is the only way. What’s the most common way people making the switch?
this is not the only way but the most common way. or at least master of ml