Kelsey Hightower is a former Google Distinguished Engineer, open-source contributor and co-author of Kubernetes: Up and Running. These lessons cover his approach to engineering, learning, leadership and life beyond work. — Screaming in the Cloud #492.

Visual summary of operating lessons from Kelsey Hightower.

Part 1: On Mindset and Life

  1. On Purpose: A meaningful life involves more than earning and exchanging money; consider how your choices help other people. — Engineering with Empathy.
  2. On Retirement: Retirement brings its own learning curve: deciding how to spend time when an employer no longer fills the calendar. — Screaming in the Cloud #492.
  3. On Gratitude: Appreciate being alive, even when circumstances could be better. — Observy McObservface #6.
  4. On Patience: Give yourself time to understand complex systems rather than expecting expertise immediately. — Humans+Tech Interview.
  5. On Humility: Stay humble enough to become a beginner again as technology changes. — HashiCorp Fireside Chat.
  6. On Identity: Do not make a particular technology your identity: the work can change while you remain. — DevRelCon San Francisco 2019.
  7. On Minimalism: Hightower uses minimalism to focus time and attention on what he considers meaningful. — HashiCorp Fireside Chat.
  8. On Success: Share what you can and learn to celebrate other people’s success without needing a personal reward. — Humans+Tech Interview.
  9. On Pace: Slow down enough to build durable understanding rather than demanding immediate mastery. — Tech Lead Journal #18.

Part 2: On People and Leadership

  1. On Core Skills: Treat people with respect by default; human skills deserve serious attention alongside technical skills. — Humans+Tech Interview.
  2. On Impact: Understand how your work contributes to the organization’s purpose rather than confusing activity with useful impact. — Engineering with Empathy.
  3. On Credit: Acknowledge the people whose support made your achievements possible. — HashiCorp Fireside Chat.
  4. On Disagreement: Treat corrections as a chance to improve your understanding rather than presenting every initial idea as authoritative. — The ReadME Podcast — Kelsey Hightower.
  5. On the Senior Human: Technical seniority should include how you treat people, not only what you can make machines do. — Engineering with Empathy.
  6. On Design: Listen and work through the problem together before arriving with a solution to impose. — DevRelCon San Francisco 2019.
  7. On Judgment: Experience helps you judge where AI tools save time and where they do not; working output alone is not enough. — Beyond the Noise #11.
  8. On Multipliers: Leadership includes helping others solve problems themselves and become more capable, rather than doing everything alone. — Engineering with Empathy.
  9. On Learning Out Loud: Ask for help rather than pretending to know; an honest question can save time otherwise lost struggling alone. — HashiCorp Fireside Chat.

Part 3: On Open Source and Community

  1. On Collaboration: Open source lets people collaborate on shared problems while using forks to explore disagreements. — Humans+Tech Interview.
  2. On Maintenance: Maintainers cannot fulfill every request. Set a sustainable scope and communicate what the project will not do. — The ReadME Podcast — Kelsey Hightower.
  3. On Communication: Explain why an engineering choice makes sense, not merely how to make the tool work. — Beyond the Noise #11.
  4. On Meritocracy: Hightower describes contributing to Kubernetes without a degree check; his work provided a way to demonstrate ability. — The New Stack Makers — Open Source’s Evolution.
  5. On Origins: Hightower sees open source’s resistance to powerful vendors as a reason to question vendors’ motives when they participate. — The New Stack Makers — Open Source’s Evolution.
  6. On Sovereignty: Accessible open-source tools gave Hightower a way to learn and contribute without first obtaining an employer’s permission. — The New Stack Makers — Open Source’s Evolution.
  7. On Open Licensing: An open-source business needs something customers value enough to pay for, such as maintained releases or support. — bynd code #8.
  8. On Empathy for Maintainers: Consider both the new user’s uncertainty and the maintainer’s workload; ask clearly, share what you tried, and avoid hostile exchanges. — DevRelCon San Francisco 2019.
  9. On Influence: Repeated requests from customers can influence a cloud provider’s product priorities. — The New Stack Makers — Open Source’s Evolution.

Part 4: On Kubernetes and Cloud Native

  1. On The Endgame: Kubernetes provides machinery for building higher-level platforms; the platform’s own controller must still do the work. — Tech Lead Journal #18.
  2. On The Future: Hightower hopes better ideas eventually supersede Kubernetes; he does not suggest it is about to disappear. — The ReadME Podcast — Kubernetes’ Future.
  3. On Abstraction: Kubernetes can make groups of virtual machines easier to manage without making the underlying machines disappear. — Tech Lead Journal #18.
  4. On Overuse: Do not introduce Kubernetes merely to process a file or develop an application locally when a simpler environment fits. — The Podlets #7.
  5. On The Hard Way: Use manual bootstrapping as an exercise in understanding Kubernetes components, rather than as a production setup or a requirement for every developer. — Kubernetes the Hard Way.
  6. On Custom Resource Definitions: Use custom resources where a declarative model fits; not every action belongs behind a Kubernetes API. — Go Time #114.
  7. On Legacy Patterns: Adopting a new platform while insisting it reproduce every old workflow can preserve the complexity you hoped to leave behind. — Stack Overflow Podcast #279.
  8. On The Cloud: Cloud platforms involve tradeoffs. Choose what fits the workload rather than treating one platform as the answer to everything. — GitLab TechExplorers Interview.
  9. On the Goal of Cloud: Offer ways to consume infrastructure without first building it, while recognizing that some users need deeper control. — Break Things on Purpose #10.

Part 5: On Engineering and Complexity

  1. On Automation: Understand the process first, then decide what automation your team actually needs. — Automation Is the Serialization of Understanding.
  2. On Progress: More code is not necessarily better software. Understanding the problem may reveal a smaller change—or nothing to build. — Beyond the Noise #11.
  3. On Tool Mastery: Know a tool’s limitations as well as its strengths before choosing it for a problem. — Engineering with Empathy.
  4. On Complexity: Do not copy a large company’s elaborate architecture without understanding the constraints that produced it. — Screaming in the Cloud #492.
  5. On Infrastructure as Code: Infrastructure changes can have side effects beyond the configuration in Git. Account for the provider’s actual state and defaults. — Platform Engineering Podcast #39.

Part 6: On Career and Experience

  1. On CS Degrees: Structured education can be valuable, but completing a degree does not remove the need to keep learning. — Humans+Tech Interview.
  2. On Experience Breadth: Do not confuse years on the job with continued growth; seek opportunities to learn something new. — HashiCorp Fireside Chat.
  3. On Entrepreneurship: Hightower’s experience running a business helped him understand employers’ problems rather than focusing only on pleasing a boss. — HashiCorp Fireside Chat.
  4. On Self-Taught Reality: Hightower describes self-directed learning as a familiar route into technology, not an inferior exception. — Tech Lead Journal #180.
  5. On Fundamentals: Learn the underlying concepts so a new tool does not feel like starting from zero. — Tech Lead Journal #18.
  6. On Career Sovereignty: Keep your judgment and identity distinct from your employer’s position; employers and roles can change. — Screaming in the Cloud #492.
  7. On Opportunity: Look for hands-on work that develops your skills, rather than relying only on an employer’s name. — HashiCorp Fireside Chat.

Part 7: On Systems and Operations

  1. On Culture vs. Scripts: Choose an interface that fits the people using it; a ticket can front an automated workflow rather than become manual work. — Platform Engineering Podcast #39.
  2. On Microservices: Separate services introduce deployment and coordination costs. Start with a disciplined monolith when it fits, then split where needed. — Tech Lead Journal #18.
  3. On Operations: Reliable operations need application cooperation, including useful health checks and sensible behavior when dependencies fail. — CROZ Interview with Kelsey Hightower.
  4. On System Administration: System-administration experience provides a useful foundation for Kubernetes. Take time to connect familiar concepts to the new system. — Break Things on Purpose #10.
  5. On Service Mesh: Service-mesh components can add deployment and configuration work; assess that overhead rather than assuming more components always help. — Tech Lead Journal #18.
  6. On Free Software: If you choose to maintain and support free software yourself, account for the work that responsibility requires. — bynd code #8.
  7. On Containers: When a container platform fails, look beneath it: the problem may be in Linux or networking rather than Kubernetes itself. — CROZ Interview with Kelsey Hightower.

Part 8: On Product and Business

  1. On Hello Revenue: Show customers how technology changes their business, not just that a demonstration runs. — DevRelCon San Francisco 2019.
  2. On Ideas: Share a tangible prototype to communicate an idea, then let others develop it beyond your starting point. — The ReadME Podcast — Kelsey Hightower.
  3. On Hype vs. Usage: Pay attention to what people actually run in production, not just what attracts excitement. — Go Time #114.
  4. On Business Context: Prepare for executive conversations by understanding the business and its priorities as well as the technology. — DevRelCon San Francisco 2019.
  5. On Breaking Things: Address recurring operational problems instead of normalizing repeated emergency fixes. — Engineering with Empathy.
  6. On the Monopoly of Power: Hightower worries that the capital and infrastructure needed to build major AI models restrict who can compete, even with several providers. — Beyond the Noise #11.
  7. On the NoCode Movement: The No Code repository is satire: its empty examples joke about avoiding software problems by writing and deploying nothing. — No Code — Satirical Repository.
  8. On Selling Complexity: Features added for individual buyers can make the overall product harder for others to navigate. — Screaming in the Cloud #492.
  9. On The Value of Boring: Successful infrastructure can become unremarkable: people rely on it without constantly thinking about it. — Screaming in the Cloud #492.