Context
Since 2024 I have been the engineering lead for several product teams at an ed-tech company, and I still write code.
The problem
The teams needed planning, delivery governance, technical direction and release ownership. Delivery was hard to predict, and review gates were often bypassed.
What I did
I run sprint governance with scripts and Claude skills that draft kickoff, demo, closure and quarterly reports.
I designed delivery analytics that measure the system and never rank people.
I integrate and review code in the main repositories. I merged 340 pull requests in the learner monorepo and reviewed at least 55 in the Rails backend, and I wrote the onboarding guide for new engineers there.
I lead audits that set direction: a 94-repo agent-readiness audit, a cross-service security audit and a mobile architecture audit, each ending in a plan.
I am a code owner in the main repositories and I own releases.
Key decision
I chose to measure delivery as a system and to keep the evidence in documents and data, not opinion. Team metrics never rank individuals, which costs me an easy way to point at one person.
I also admit that my own integrator role partly worked around review gates. That is what pushed me toward governed agent delivery.
Result
Sprint reports have run every sprint in the second to fourth quarters of 2026. Team features shipped through the integration role.
What I learned
I would again lead through audits and written plans. Next I would build review gates that do not depend on one integrator.
Stack
- Python
- GitHub
- Jira
- Confluence
- Claude Code
