Decision
In this project, I made a clear decision early on when faced with the option to tell another team to do a database migration or drop a dashboard customers found useful: “Drop the dashboard. The product is causing these users to do more work. And don't migrate the database. The data schemas are correct. I have found a way for our operations teams to provide this dashboard data and provide more holistic support.”
Context
In 2025, I took the lead to bring more cohesion across Partner and Merchant Engineering at Affirm through a merchant-domain architecture review that set which team owns which merchant data.
How I led it
As the engineering manager overseeing this program, I set up working meetings with every Partner and Merchant Engineering team to enable our tech lead to see around architectural corners only the other tech leads would know about (e.g., are we choosing to consolidate this endpoint or will it become deprecated in a future build?). I also set up kickoff and close meetings with upward and peer leadership to say, “Work with us! And if you are too busy, tell us.”
Result
We got sign-off from 2 Directors, the Principal Architect, and every team's EM, and I presented it to Affirm leadership in Q3 2025. The review showed Merchant Risk and Merchant Advocacy were tightly coupled, which led Affirm to consolidate them under one EM, move former developer-support engineers into SRE, and add PM investment in direct merchant customer problems.
Belief
Building system architecture is too often seen as simply a requirement to appease the high standards of enterprise customers. System architecture, along with any technical debt work, is always an act of building the core product.
“Are our systems fast enough to get hesitant users to think, 'that was easy' and go tell their friends?” “By reducing confusion about data discrepancies in our system, will we grow the bottom line?” “If we have another incident, will users leave us, or will they leave us because they just didn't understand how valuable the dashboard was to them?”
Needing operational support isn't a failure, even at a company led by product and engineering. It's how a system grows into something that feels human.