Meet Katie Baker, Tribe ML engineer and data scientist:
  • From photographer to data scientist – hear how Katie fell in love with data, spent 10 years using ML to solve problems in healthcare, pivoted to the startup life, and then turned to data consulting to take back ownership of her time.

You came to data science and ML by way of photography. I love an unusual origin story! Tell us more.

I’ve been in DS and ML roles for ten years now, but my degree is actually in photography. I always needed a mix of technical and creative – I was pursuing math and then I switched to art and eventually ended up in photography, which I think is one of the more technical art forms there is.

I got a job doing entry level data ingesting in healthcare to support my plans for a photography studio and just fell in love with data. From there I taught myself SQL and Python and during that time I was taking courses on data mining and statistics and connecting it all back to the mathematics I’d loved in the first half of my college career. In the end, I spent the next seven years there as a data scientist.

What were some of the initial problems that got you hooked on data?

We had this big problem at the time, which is that healthcare was moving from one codification system to an updated one, which was a huge shift in healthcare claims data. So many of the models in the space were based on the first coding system, so it broke how we were risk adjusting patients for mortality, for readmission, essentially for everything. So we had to brainstorm a solution for this and we prototyped a Bayesian model in SQL that we ended up patenting.

We were coding in Java at the time, a requirement based on the company’s tech stack, so we were implementing neural networks from scratch, all of these things without the use of packages. I found I just loved software engineering. I liked the challenges that came with machine learning, especially. I liked the unique issues that come up with memory consumption and various optimizations to make things performant and how they get consumed within data pipelines and software applications.

After 7 years at an established healthcare company, you pivoted to a startup. Was that a big change?  

Healthcare is amazing – really interesting and impactful. It’s also really slow. What you can do with the data legally is really limited. The company ended up moving my team onto the same floor as the lawyers because we had to consult legal so often. So I ended up pivoting to a startup in the FinTech space. It was completely different – it was building the entire architecture from scratch instead of band-aiding old systems, it was building the team from scratch.

But at a certain point, my role became really managerial, very big picture. It was all about planning three quarters ahead. I liked mentoring and building a team, but I missed building something new. That’s what made me start thinking about consulting.

You consult solo in addition to consulting for Tribe. What feels different?

When you’re generating projects from your own network, there’s a certain bias towards industries you’ve worked with in the past. So working with Tribe, you get exposure to companies you never would have interfaced with. Right now, I’m working on a Tribe project that’s in healthcare but it’s very different than other problems I’ve worked on in this space before. The data is more holistic than claims data and the company is completely data driven in a way that feels really rare.

So what does the future hold?

It’s hard to predict, but the types of projects I do as a consultant feel much more inline with what I want to do. I’d be hard-pressed to say I’d ever go full time anywhere again.

I don’t want any one thing to consume all of my time. For me, it’s about variety. I want to work on something new every day. I want to meet interesting people. I want time to pursue other passions – photography, family. I want to balance out screen time with time physically in the world. I’ve also considered becoming a doula – I’m really passionate about women’s health. To me it’s about time. Ownership of my time.