Descriptions:
Leo Mehr, Director of Engineering at Ramp, shares two core principles his forward deployed engineering team has developed over two and a half years — scaling from two engineers to a 30-person organization covering deployed developer API and AI services. The talk, delivered at the AI Engineer World’s Fair, is deliberately focused: two lessons, illustrated with specific stories from Ramp’s enterprise customer work.
The first principle is disciplined scoping. Mehr argues that FDEs instinctively default to saying yes to every customer request, an impulse that produces waste. He illustrates the cost with a painful early example: his team spent weeks building a mobile reimbursement feature for both iOS and Android, only to discover at delivery that the enterprise customer mandated iOS-only devices company-wide. Proper scoping — interrogating urgency, exploring workarounds, checking whether other customers have the same need — is what separates impactful FDE work from expensive busy work. The principle extends to resisting sales-driven urgency at quarter-end that doesn’t reflect genuine customer need.
The second principle is scaling with tokens: systematically replacing each stage of the FDE workflow with agents and LLM-assisted tooling. Mehr demonstrates this with Ramp’s internal FDE request pipeline, where incoming customer blocker requests posted to a Slack channel trigger an agent that reads the linked Notion document, assesses what exists in the product, and produces a structured triage — work that previously required manual FDE review. He frames this as the essential survival skill for the role: engineers who don’t automate their own FDE work will fall behind those who do.
📺 Source: AI Engineer · Published July 28, 2026
🏷️ Format: Keynote Launch
