Am I planning it right for a PhD in ML?

I am a Master's student in EE from a pretty decent school, specializing in Image Processing and Machine Learning. I want to do a PhD in the long run, but I want to be fully prepared, armed with a solid knowledge of the basics and good experience in the field before taking my 5-year plunge. Basically, when I start my PhD, I want to hit the ground running, doing research, and not waste the first year just studying the basics.

Toward this end, my plan after my Master's is to do a research internship in some lab for a year and hope to get a/some publications out.

I just want to ask you guys, will this really help my PhD applications? I want to get into a good PhD program.

What about working in a ML startup? I know there are a LOT of companies out there with TONS of data, will it boost my resume if I worked in some such core ML company (even if only a small startup), gained good knowledge, but got no publications?

Thanks a lot!