Surprises in becoming an engineering manager

August 4, 2022

Recently I talked to a few folks that were curious about transitioning into software engineering management, and we talked about what it was like when I made that transition about 4 years ago. I’ll summarize most of those conversations in this post. I’ll focus on the surprises I found in the first few months of management.

Before jumping into what I found surprising, some context first: My former position might best be called lead machine learning engineer – I was an individual contributor spending about 20 hours per week on leadership efforts like mentoring and the roadmap in addition to research and development. I transitioned to manage a team of 6, which was a mixture of machine learning engineers and software engineers.

Surprises

The workload was much heavier as a manager, both in terms of hours worked and stress. Over the years, I got better at the job and it returned to normal levels. I also later realized that it was partly self-inflicted: I cared a lot about my team and also cared a lot about doing a good job, so I pushed myself hard.

There’s less mentoring and training for managers than individual contributors. I found that shocking, given the outsized positive or negative effect a manager can have. This seemed to improve as the company matured and larger companies tend to have more established programs for this.

It was much more rare to accomplish everything I wanted for the week. I had to re-calibrate to set realistic goals for myself. Also, it was harder to break goals down into smaller steps. For example, I might want to improve one person’s ability to factor code into smaller, independent pieces. That’s not specific enough to even say when it’s done, and even if it were, the solution could range from a simple conversation to months of trying different coaching strategies.

Management wasn’t initially a good financial decision. I was doing much more work and earning only slightly more, so my hourly wage had effectively decreased. I had known about the philosophy that management was a parallel track to IC work, but didn’t think about it enough to understand the implications. It took years to reach a compensation level that felt right for the level of responsibility. I’d expect this to vary by employer.

I had less influence over the overall product development roadmap than I expected. I naively thought that the director title would grant me more influence, but that was rarely the case with other leaders.

I wasn’t sure if I’d like management or not, but I expected that I’d try it for a while and learn that I loved it or hated it. I was wrong on that count: There were times when I loved it, and times when I hated it.

Management was more of an emotional rollercoaster. I could have a wonderful morning like interesting feedback from our users then a miserable afternoon like an employee worried about being kicked out of the country due to visa issues.

Even experienced engineering managers had little familiarity with machine learning, so I spent a fair amount of time teaching peers. There’s also much less advice out there for managing hybrid research and development teams, in contrast to management of more traditional engineering disciplines.

It’s tougher to explain what a typical day is like as a manager. It’s a wide range of things, from technical leadership to product leadership to routing information to seeking information to connecting people to interviewing and so on. The tasks really depend on what the team needs most, which depends on the day. That said, the core job is pretty clear: You’re accountable for the total productivity of the team against the business’ needs.

Finally, management felt much more entangled with my personality, for example I prefer to be candid even when things are going badly. If that’s communicated well, it can earn the trust of the team. If that’s communicated poorly, it can distract people or even lead to attrition. I don’t have the best words to describe it, but I felt much that it took much more effort to be authentic as a manager.

Non-surprises

• It’s a meeting-heavy workload.
• I had to write more documents – I vaguely knew about this, though I hadn’t thought about it in detail. The documents could range from levelling documents to team charters to quarterly roadmaps to multi-year vision statements to investor/customer inquiries.
• There’s tons of management advice out there, but almost none of it is backed by evidence – I knew about this somewhat already but it didn’t fully hit me until years later.
• It took more effort to have candid conversations with my team.

Hope this helps some of you that are curious about management!