“The highest-functioning teams I've seen run on clear guardrails and real autonomy. I keep an airing of grievances slot in my meetings so people argue openly, then walk out unified. The CTO job is mostly making sure everyone who reports to you is better than you at their job and getting out of their way.”
“I stopped thinking of engineering teams as engineering-driven or business-driven; the ones that last are people-driven. We do not exist on islands receiving orders and delivering outcomes. We exist together, supporting, leveling up, trusting, and building for each other. Optimizing for the smaller part has never been the path to winning; optimizing for the whole is what gets a team where it needs to go.”
“The highest-functioning team I ran was not the one with the best engineers; it was the one where people were not afraid to ship. At CashApp I watched a 4-person team grow to 50, and the thing that broke first was always the deployment and review culture, not the code.”
“I spent years shielding my team from every hard conversation, thinking that was leadership, and it backfired: they stopped owning outcomes and I became the bottleneck. Now I hire for comfort with ambiguity and low ego, because great teams build great software, not individuals.”
“A high-functioning engineering team, in my experience, is one where the business side and the developers stop talking past each other. At SiteOwl, the time zone between the US and India became an advantage once requirements were clear enough that problems reported in the US morning were fixed by the time the US team woke up.”
“I scaled an engineering team from four to fifty-plus at Endurance, and the lesson that stuck was that the best ideas came from cross-functional trust, not the smartest engineer in the room. My job as CTO was to get people aligned, then get out of their way.”
“I have run teams from five people to 150, and the thing that changes at scale is not process; it is trust. At Ring I watched a 2.3 app rating climb to 4.8 once the teams believed quality was theirs to own, not a gate above them. At Tala we cut deploys from weeks to minutes by making the platform so observable and safe that engineers stopped fearing release day.”
“I spent too long watching teams slow down because every deployment felt like a coin flip. The biggest unlock was decoupling the architecture so engineers could push code without fearing they'd break something three services away. Psychological safety isn't a soft metric; it's the thing that lets velocity actually compound.”