# What Changed When I Started Leading Engineers

*Published 2024-06-02 · Leadership, Career*

> Going from senior developer to managing 12+ engineers rewired how I think about code reviews, estimates, and the quiet work of unblocking people. Lessons from the transition.

Source: https://www.yashthakur.com/blog/leading-engineering-teams

---

**The short answer:** the thing that actually changed when I moved from senior developer to leading 12+ engineers was realizing the job is a different discipline, not a scaled-up version of the old one — my output became my team's output, code reviews shifted from correcting to teaching judgment, estimates became conversations about uncertainty rather than numbers, and most of my highest-leverage work turned into the invisible job of protecting people's focus and unblocking them.

The first time I was handed a team, I made the classic mistake: I treated management as "senior engineering, but with more meetings." I kept reviewing every PR line by line, kept writing the trickiest code myself, kept being the bottleneck I was supposed to remove. It took a couple of painful quarters at Maxxton — growing from team lead to associate technology manager over a team of 12+ — to understand that the job had actually changed, not just expanded.

## Your output is now other people's output

As an IC, a good week is measured in features shipped. As a lead, a good week is measured in features your *team* shipped and the obstacles you cleared before they became blockers. This is genuinely hard to internalize, because the feedback loop is slower and less visible. You can finish a day having written zero code and still have been enormously productive — if you unblocked three people, prevented one architectural wrong turn, and gave a junior engineer the context to make a decision themselves next time.

## Review for teaching, not for correctness alone

Early on, my code reviews were a list of things to fix. Effective, but it meant the same comments appeared on the next PR. I started reviewing differently: fix the blocking issues, but for everything else, explain the *why* and the tradeoff, and let the author decide. A review that says "this works, but here's what happens to it under concurrent writes — your call" teaches judgment. A review that says "change this" teaches obedience. One of those scales; the other makes you the permanent bottleneck.

I also learned to approve more. A PR that's 90% of the way there and unblocks someone is usually better merged with a follow-up ticket than held hostage over style. Perfect is the enemy of a team that's moving.

## Estimates are a conversation, not a number

The most useful thing I did for delivery was to stop treating estimates as commitments and start treating them as a shared understanding of uncertainty. "Two weeks, assuming the payments API behaves and we don't discover the legacy migration is load-bearing" is infinitely more useful to a product partner than "two weeks." It names the risks out loud, which means they get managed instead of discovered at the deadline.

## Protect focus ruthlessly

The thing engineers need most and get least is uninterrupted time. A lot of management is absorbing interruptions so your team doesn't have to — fielding the stakeholder question, triaging the "quick favor," batching the status updates. I became the shock absorber. When I did it well, the team barely noticed the chaos I was buffering, which is exactly the point. Invisible work that keeps good people in flow is some of the highest-leverage work there is.

## You will miss the code — do something about it

Nobody warns you how much you'll miss building. I keep a foot in it deliberately: side projects, this very portfolio, the occasional gnarly bug I pair on. Not because the team needs me to, but because staying close to the craft keeps my reviews grounded and my estimates honest. A manager who hasn't shipped in two years slowly loses the ability to tell a real two-day task from a real two-week one — and that erosion is quiet until it isn't.

The transition from engineer to leader isn't a promotion so much as a career change that happens to share a job title. The sooner you treat it as a new discipline to learn rather than your old one scaled up, the sooner your team gets a manager instead of a very distracted senior developer.

The groundwork for this starts before the title does — [the senior engineer skills nobody teaches you](/blog/career-growth-senior-engineer) is the judgment layer that makes the jump survivable — and if your team is distributed, [running a remote engineering team without the burnout](/blog/remote-work-engineering-team) is where most of these habits get tested hardest.
