Fixing the codebase will reveal we are massively overstaffed. What to do?
retained text
I'm a Principal Engineer leading a team of about 10 engineers at a Series D startup. I've been in the role for a little over a year, and now that I'm settled in, I'm looking closely at how we can improve our engineering velocity and code quality. Right now, a large portion of our sprint capacity is spent on churn, mostly fixing regressions introduced by recent PRs. Because everyone is leaning heavily on AI coding tools without standardized guardrails or thorough code reviews, a lot of unvetted code is making it into production which has led to significant technical debt. Most of the engineers have zero concern about code quality, or even know what good code looks like. I have a clear path to fix this. I could clean up the core architecture, set up automated guardrails, and configure standardized agent skills to make feature development much smoother and more reliable. I've already rolled out this setup in a few isolated repos with great results. In the repos this team isn't responsible for, my job is easy. Feed the agent a ticket, come back a few hours later to verified results. Meanwhile, they are opening countless PRs fixing the same issues over and over. Once the codebase is removed of the tech debt, and the tooling is in place, our velocity will increase dramatically, to the point where a team of 4 or less could easily handle our current roadmap. I want to do what's best for the product and the company, but I'm worried that highlighting this efficiency gap will inadvertently put the rest of the team's roles at risk or make it look like the team is heavily overstaffed.
document fetched 2026-11-16 00:23 · canonical https://reddit.com/r/cscareerquestions/comments/1xx8ven/fixing_the_codebase_will_reveal_we_are_massively