Atlassian

There are some smart and kind people there.

Incredibly high turnover. A majority of my team had less than 6 months' experience. You'd think they'd pair new people with experienced people, but no, they'd put all the new hires together.

Your manager would ask you to do something, then during your review, they'd tell you that the thing they asked you to do lacked business impact. My manager flat-out told me that some teams will have more business impact, and that the team I was on was maxed out at "meets expectations" no matter how good we were.

My face time with my managers ranged between 2-6 hours a quarter.

You will always have to work outside of normal business hours. I had scrum calls at 11:30 at night. I got no help from our platform team during my working hours, which made my tickets stretch on or forced me to work until 2:00 am to be online at the same time as them.

The managers I encountered looked at everything so simplistically, with barely a surface-level understanding. This is the level of thinking I encountered: "Person A made 5 tables, Person B made 7 tables. Therefore, Person B > Person A." Zero understanding of variations in complexity of tasks.

That same shallow thinking applies to PR (pull request) count. If Person A thoroughly tests and validates their code and deploys once versus Person B, who releases a buggy piece of crap and has to deploy 10 patches, management will applaud Person B. If Person A writes their code DRY and Person B is a copy-pasta wizard, management will applaud Person B.

You need to dump or significantly devalue the PR count and line count metrics. They are so easy to game; they reward releasing bad code and fixing it later, they punish testing and validation, and they devalue all of the other things that go into data engineering like requirements gathering, planning, and data analysis. (I saw some PRs that merely moved a bunch of JSON data from one file to another but were credited with 5,000+ lines of code written. SMH)

Stop holding data engineers accountable for doing things they are asked of by their managers and TPMs that don't add value. For example, if I'm asked to build a data pipeline or run analysis to detect fraud, and if it finds $1 million of fraud or $100 of fraud, that's the same amount of work for me. So why would I be punished or rewarded based on that outcome?

Managers and TPMs should be a little more hands-on. How are they supposed to rate someone that they only talk to less than 1 hour a month? That's how you get managers with a surface-level, child-like understanding of what their team does. That's how you get the "More PR = GOOD, More tickets = GOOD" mentality that just drives people to spend their time leaving a trail of forensics to create the illusion of work rather than reward the people that do real work.

Use our scale effectively. I'm still in awe of how foolish it was to deploy hundreds of data engineers to manually trace their data pipelines, find their dependencies, and find their users, etc., to plan migration. All of the data needed to answer those questions existed in logs. A small team of people could have automated the parsing of those logs and saved huge amounts of time. I suggested automating that, among other things, but was always shut down and chastised when I tried to solve things at scale.

How are we supposed to "Play as Team" when everyone is pitted against each other through stack ranking?

To summarize, it feels like no one in management has heard of game theory. So maybe read up on it, then rethink your policies based on what you learn.