← Back to blog

Rediscovering my Craft

Rediscovering my Craft

The Forced Reset

In early 2025 I took medical leave. I won't sugar coat it. I was burned out from work stress, the world, and my wife's disability, and it came with real physical symptoms. I went cold turkey on tech to see what would happen.

At first, I didn't miss it. I rode my bike, wrote my novel, slept, ate better, and wrestled with the American obsession with productivity. We split time in Portugal, which made it easier to step off the Silicon Valley treadmill. Eventually, I found a rhythm.

As my return neared, I eased back in: industry reading, catching up with old networks. What I saw was familiar. AI is reshaping things, sure, but the hard problems hadn't moved: production outages from schema changes, slow queries on the wrong indexes, databases treated like black boxes. Now add LLM/agent chaos layered on shaky fundamentals. Different tools, same gravity. And the spark came back.

What Still Matters (and What's New)

My core is databases and data infrastructure. I've done this at scale—from building Travelocity's first DBA team and running PalominoDB to co-authoring Database Reliability Engineering and leading production engineering for online data at Meta. That expertise is still relevant.

The principles haven't changed: ACID, replication, query optimization, capacity planning, backup and recovery. The venue has: Kubernetes pods instead of bare metal, Terraform instead of hand-rolled provisioning, streaming stacks integrated into the data plane. I don't want to be a generalist infra engineer; I want to be a database specialist fluent in modern environments.

Why the Pendulum Is Swinging

The executive track is comfortable. But I missed the work. I wanted to write IaC instead of just reviewing diagrams, to debug live issues and land big wins from the keyboard—not read about them in the postmortem. After a decade leading orgs, I'm swinging back toward building.

This isn't a retreat from leadership. It's a deliberate move toward hands-on credibility while keeping the strategic lens. Think Will Larson's staff-plus continuum, TL/Manager hybrids, and Charity Majors' IC/manager oscillation. Careers aren't ladders; they're pendulums.

Why It Matters

Better decisions. Strategy untethered from implementation drifts into wishful thinking. Staying close to the work keeps leaders honest.

Credibility. Engineers can tell who's current. Respect earned in 2015 has a half-life. Shipping last week matters.

Growth. I've optimized org performance for years. Useful, and I'm proud of it. But the tools moved on. If I want to be effective for the next decade, I need new calluses.

Joy. Strategy is satisfying. Mentorship is meaningful. But there's a specific pleasure in making a system work with your own hands.

What Success Looks Like

I'm not becoming a full-stack generalist. I'm doubling down on databases and data systems, applied with today's toolchain.

Shape of work: Principal-level database reliability + technical advisory. Walk in at multiple layers: talk strategy and cost with the CTO, design architecture with the VPE, tune replication and queries with the DB team, then implement via their Kubernetes + Terraform stack and wire up monitoring.

Hands-on ratio: Roughly 60% deep technical (databases/data infra), 40% strategy/leadership/advising. Ratio flexes by engagement; focus doesn't.

The Learning Plan (and the Privilege to Execute It)

I can do this because I have Meta-level savings. I don't need to maximize short-term income. That's a real privilege, and I'm not prescribing it to anyone else. My goal is to document the path so others can move faster with less risk.

Time budget: 15–20 hours/week of intentional, hands-on learning and shipping. I may be underestimating; that's fine. I'm starting with deep systems knowledge and mapping it to new paradigms.

Three parallel tracks:

Modern database paradigms, not Kubernetes cosplay

  • StatefulSets for PostgreSQL/MySQL and the HA tradeoffs that actually matter.
  • Secrets management for DB credentials; persistent volumes; backup/restore you'd trust at 3 a.m.
  • Terraform for right-sizing, cost controls, and repeatable database provisioning.
  • Streaming: Kafka/Flink and newer options where they make sense—not because they're shiny, but because the workload demands it.

AI-assisted coding, used critically

  • Copilot/Cursor/Claude as accelerators, not crutches.
  • Describe the desired deployment; get YAML/Terraform; read, verify, modify, break, fix. Internalize patterns; don't outsource understanding.

Build real things

  • Opinionated, practical tools for DB monitoring/alerting I've wanted for years.
  • Reference implementations: HA Postgres on K8s with sane backups and failover; replication-lag detection that pages before customers tweet; streaming pipelines that survive reality.
  • Write it up: posts, playbooks, maybe a refreshed Database Reliability Engineering second edition that reflects current practice.

Reality check: I may misjudge the gap or the time. Some folks will want recent proof, not history. That's fair. The downside for me is discomfort and recalibration; the upside is relevance and impact.

Where I'll Add Immediate Value

  • Help where AI meets databases. When vector DBs buckle under weird query shapes and concurrency spikes, I want to read the code and configs, fix anti-patterns, and prove the fix in their environment.

  • Database replication that stays up. Pair with their team, land the Terraform changes, set up monitoring for lag, and harden failover.

  • Query performance that holds under load. Identify the 5% of queries causing 80% of pain and fix them with indexes, plan stability, and sane limits.

  • Deep investigations that yield immediate and valuable improvements.

An Invitation

If you're a leader who's drifted from the work and miss building, you're not alone. The risk of staying purely managerial is slow obsolescence. Not everyone can afford a pivot. If I can map this transition, I hope it shortens the route for others.

If you're earlier in your career: don't let the pendulum swing so far you can't feel the pull back. Stay a little hands-on, always.

I'm excited again: shipping code, debugging production, learning the current stack, and operating at the intersection of strategy and implementation. If you're on a similar path, I'd love to hear from you. I'll be sharing at https://appendonlylog.ai/, and you can reach me at laine@appendonlylog.ai. LinkedIn also works!