Setting expectations right - Applied Science at FAANG as a fresher would be kind of tough to get.
Most people without PhDs are given MLE roles, for FAANG in terms of qualifications it usually entails a few publications in top conferences. I have seen people get it without publications as well, but the research path is almost a sure shot way of getting interview calls from these places.
If you are really serious about it, you can apply for Googles pre-doc research engineer positions, or research fellow at Mfst research.
Long term, I would suggest try to get associated with a lab in a good university and contribute to research.
An even lower hanging fruit would be contributing to ML open source repositories.
In terms of skills, build mastery over the internals of ML models, understand how neural nets work. Then you can move on to specific architectures. Stanford 224N is a great resource.