Roy
Rapoport

When Roy Rapoport walked into Netflix in June 2009, he was unemployed, mildly anxious, and certain he'd be fired within a year. His new role was redefined twice in his first week. What followed was eight years and seven months of building something genuinely rare: an engineering team that nobody wanted to leave.

Roy's world is the operational layer of technology - the invisible machinery that keeps services running at scale. At Netflix, he ran Insight Engineering, the group responsible for cloud telemetry, alerting, and real-time analytics. You may not know the name, but you've benefited from the work every time Netflix loaded without drama at 8pm on a Friday. He also helped build the CORE group (Cloud Operations and Reliability Engineering) and wrote most of Netflix's Python infrastructure libraries that let developers talk to cloud systems without reinventing plumbing on every project.

That's the job description. The actual accomplishment is harder to quantify: over three and a half years managing Insight Engineering, he had zero voluntary departures. In an industry where good engineers get recruited every quarter, zero is a number that makes other managers sit down and ask questions.

Roy has a second career running quietly alongside the engineering work: he writes. Not the kind of writing that teaches you to think in buzzwords. The kind that makes you stop and realize you've been thinking about management wrong for years.

His "Five Conditions for Improvement" is one of those frameworks that sounds obvious once you read it, which is a sign it's actually correct. The core argument: before any employee can fix a performance problem, five things need to be true. They must acknowledge the problem exists. They must want to fix it. They must see their own role in it. They must be capable of forming a plan. And they must be able to execute that plan. Skip condition one and condition five becomes theater. Roy's insight is that most managers sprint straight to "here's your plan" while the employee hasn't even reached "here's the problem I acknowledge having."

His "Manager Impact on Report Performance" piece tackles something almost nobody talks about directly: the same employee can flounder under one manager and thrive under another. This isn't a comfortable observation for most organizations, which prefer to locate problems entirely within individuals. Roy locates them accurately.

Then there's his series "Distinctions and Concepts" - a collection of compressed, precise ideas about how communication and trust actually function in organizations. "Language as Variable Passing" treats words the way a programmer treats memory: imprecisely named variables cause bugs. "Forms of Trust" dissects the different things we mean when we say we trust someone, and why conflating them causes real damage. These pieces are short. They stay with you.

In 2016, Roy wrote a document about himself for his team. Not a performance plan or a vision statement - a plain description of who he is, how he works, what he values, what his quirks are, where he fails. He called it his Manager README, and he posted it publicly.

The industry picked it up immediately. Within months, engineering managers everywhere were writing their own. Today there are hundreds of them floating around GitHub, Medium, and company wikis. Most of the current wave traces directly or indirectly back to Roy's document. The concept wasn't entirely new - Luc Levesque had documented something similar in 2012 - but Roy's version hit the community at the right moment with the right level of candor, and the format spread.

He has since written a Manager README for every job he's held. He describes them as "somewhat useful" - understated, which is on brand.

In early 2018, Roy left Netflix. He counted: 3,138 days. He was emotional about it - "I love this place, and its people, and its culture. I'll miss it terribly" - and then he went to Slack.

His path to Slack has the structure of a good story. He first noticed Slack in 2016 when the company sent four Black women engineers to accept a Crunchie award. That's how an organization signals what it values. He started talking to Slack leadership. They offered him a role. He declined - his Insight Engineering team was in crisis and he refused to leave mid-emergency. Slack kept the door open. A year and a half later, while he was on parental leave after adopting his daughter, they called again. The timing worked. He went through the interview loop and noticed five of his eight interviewers were women, which he found significant enough to write about.

His first day at Slack ended with a company presentation whose entire message was: "You Belong Here." After a month, he found the message was true across every interaction. He joined as Director, Data Engineering and eventually returned to Netflix in a Director, Corporate Engineering role.

Roy lives in the Bay Area (Pacifica/Los Gatos, California). He raises goats. His Medium bio reads: "I have goats. I work in technology. You know most of the rest." It is, characteristically, both modest and exact.

He is married, and adopted a daughter around 2014-2015 while at Netflix. He is quantification-oriented - he doesn't just know that he worked at Netflix for a long time, he knows it was 3,138 days, computed to two decimal places. He tracks his own retention metrics as a management discipline.

He's been candid about having had his own performance struggles - he experienced a significant challenge in 2016 and resolved it, which informed his writing on the subject. He thinks transparency about these things is part of what makes a leader trustworthy. He appears to leave money randomly in conference rooms as a pastime, which he has declined to fully explain.

On the internet, he is active on Mastodon at @royrapoport@hachyderm.io and has 9 repositories on GitHub, including channelstats (a Python tool for Slack analytics) and cronslack (a tool for scheduling Slack messages via Slack). The GitHub bio simply reads: "Connected to Slack."

Roy's technical foundation is in operations and automation. He came up through IT engineering, spent time at Macromedia as Director of IT Engineering from 1997 to 2001, then managed back-end systems at Robert Half International before Netflix recruited him in 2009.

His technical contributions at Netflix are substantive. He built cloud telemetry systems and alerting infrastructure at a scale that very few organizations operate. His work on canary analysis - the discipline of running a small percentage of traffic through a new version of software to detect problems before full deployment - was the subject of a featured interview at QCon New York 2014. He spoke at SREcon Europe 2016 on next-generation alerting and fault detection. He's presented on Netflix's monitoring system at multiple conferences, including making the case for why most companies shouldn't build their own.

His broader view of operational capability is holistic: it's not just about the tooling, it's about the culture that decides what to measure, what to alert on, and what to do when things go wrong at 3am. He helped establish Netflix's Cloud Operations and Reliability Engineering (CORE) group and has continued in an advisory capacity after moving to other roles.