The opportunity
GitLab is the intelligent orchestration platform for DevSecOps. GitLab enables organizations to increase developer productivity, improve operational efficiency, reduce security and compliance risk, and accelerate digital transformation.
What you'll do
Define and continuously refine the operating model for the non-linear: productivity pilot: what "friction" means, how it's diagnosed, and how fixes are prioritized against the three system-level metrics.
Identify and prioritize non-linear productivity opportunities (step-change: fixes rather than incremental gains) across the SDLC, including flaky tests, review latency, pipeline reliability, tooling gaps, and context loss for both human and agentic contributors.
Translate an intentionally broad, ambiguous mandate ("find the friction, fix: it at the root") into a concrete, time-boxed roadmap with defined milestones for a 3-month pilot window.
Personally source, evaluate, and select 4 exceptional engineers for the pilot: squad. You own this hiring bar directly rather than delegating it to a pipeline.
Negotiate allocation models (100% vs. partial/dual-hat) with each engineer's: current EM, balancing team commitments against the pilot's needs.
Set the technical direction and working norms for a small, fast-moving team:: how work is scoped, how decisions get made, and how progress gets reported.
What they're looking for
- Mentor the engineers on this team in root-cause diagnosis and shift-left: thinking, raising the bar for rigor versus reaching for the nearest patch.
- Lead hands-on investigation to instrument and baseline the three top-level: metrics (MRPM, pipeline success rate, pipeline latency) so the org has an honest starting point before any fix is proposed.
- Diagnose the highest-leverage friction points dragging each metric down, and: validate hypotheses with real data before committing team time to a fix.
- Design and implement fixes that compound, including shift-left quality: changes, tooling investment, and better context and interfaces for both humans and agents, rather than one-off patches.
