This was inspired by Nicholas Zakas's and George Stocker's posts on the same topic.

Why am I doing this?

Partly, this is just selfish. If I don't actively track this, I'm never going to remember any of this information. Not that it much matters in retrospect, I'm just a data hoarder.

A little less selfishly though, nobody (in terms of folks in the workforce) benefits from secrecy in salary info. In data science and machine learning especially, I've noticed a trend where companies try and say, "Oh, well we have these existing data analyst pay bands, and data science is probably basically the same, so we'll just use those as our benchmark." It's not malicious, it's just non-obvious what the distinction always is, or what different skills and tools each job demands. But the net effect of this being roles like "Data Science Analyst" which severely underpay relative to market rates for actual data science postings. Or teams with folks who are being paid wildly different amounts solely as a function of how they negotiated their job offer. (Relatedly, go read Patrick McKenzie's article on Salary Negotiation. Seriously, go do it.)

Point being, salary transparency is pretty undisputedly a good thing. And here in the midwest, for data science roles in particular, there's just not a ton of data out there.

The data

These data were largely taken from offer letters, and are accurate to the best of my knowledge and recollection. My ending salaries are based off of memory. YOE reflects the number of years of experience I had when starting a position, and I have included any signing bonuses, year-end bonuses (as a percentage of my base salary), and stock options. I am not including other benefits, like insurance, wellness perks, employer 401k matches, and so on. These all tend to fall within a stone's throw of one another from one employer to the next, and have never been a deciding factor for me in choosing one offer over another.

[1] — Bonus target was revised from 10% to 20% at some point in my time at State Auto