This is sort of a job question thread, but the thread seems to have been largely ignored for 3 weeks so I figured I'd have better success posting a new self post.

I am a recent (1 year ago) PhD graduate from Physics (experimental, condensed matter) who has spent the last year working on my own start-up and in general staying up to date with the machine learning community and publications. I'm not sure the best way to continue from here towards building a career and figured it would be best to just ask.

My startup was based around a lot of natural language processing applied to scientific abstracts and data. However, because of the nature of startups (And doing it alone) I ended up spending some 95% of my time working on full-stack development (hosted a site on AWS with a full data pipeline, sql, message brokers, ec2 cluster scaling, etc.).

Other experience I have directly in data science come from my self-study of machine learning in general (learning the wide variety of techniques and tools available), and more recently (past 2 years) teaching and mentoring of many of my friends and colleagues in data science (some of whom have jobs doing data science now).

During my PhD I also did work in CUDA and am working on getting a patent on this for some largely parallel real-time data processing tools I created.

For my PhD in general, I worked largely and mostly in analyzing and modeling a specific type of 4D data sets, traveled around the world taking data and organizing it, and everywhere I went, automating scientific equipment (I don't like turning a knob when I can get the computer to do it).

For my undergrad I got a double degree in Physics and Electrical Engineering, with some research experience in cognitive and adaptive radio (which is somewhat related to the larger goal of getting into machine learning and data science).

All in all, I consider myself to be ready to take on almost any challenge in data science, but I am frankly a bit worried that I will be left bored and frustrated with whatever job I get if I just apply and accept whatever job offers the most income. From my friends in the field and my own experience, it seems like the meat of the work is usually just data-pipelines, scripts for data visualization, and full-stack style development: and frankly I want to avoid these things and focus on more challenging and engaging machine learning problems.

TL;DR This brings me to my final question: should I be focused on trying to just get any job, even if it isn't doing novel data science like I would like and try to build a name for myself in the community, and likely hop between jobs until I find something that fits me personally? Or, should I instead invest my time in doing direct data science myself (prototyping new architectures and bench-marking them, etc.), working on kaggles and making my attempts and code public, trying to gain notoriety that way through the scientific side? At the end of the day, the dream is to work at a company like DeepMind or OpenAI, but I recognize that my resume just isn't quite up to snuff yet and I'm wondering where I should go from here to make that happen.

Thank you for any input you may have.