I think the other commenters suggesting a better role is the best idea. Believe it or not ... there is a lot of industrial R&D ... people don't publish papers on all of it. Key outputs are Proof of concepts and patents. If you find a role like that, you will likely have the best chance to succeed.

That said, lets talk about the project idea .. here is what I suggest. Do a solid project (or three) and give a talk on it at a meet up or tech conference.

Data science/ML is stupid hot right now as an area. But the business people don't give a crap about algorithms. They want outcomes or whatnot. So, when you do a project, don't go fully deep into an algorithm or eval. Focus on a business problem (could be some trivial data analysis on open data) and write a blog post on it/do a talk. Use visualizations. use tools used in industry for this (e.g. Tableau or whatever).

Best of luck to you.