Recommends
Books, tools, and resources that changed how I think or work. No giant affiliate list. No productivity theatre.
28 references · 7 collections · maintained by hand
Start here
If you only read a few things:
- BookStaff Engineer ↗
The clearest thing written on what Staff+ actually means day-to-day.
- BookThe Staff Engineer's Path ↗
Practical and grounded. Less theory than Larson, more about navigating real orgs.
- BookDesigning Data-Intensive Applications ↗
Makes platform engineering feel like systems design, not YAML wrangling. Read it even if you don't work on databases.
- ResourceSite Reliability Engineering ↗
Useful for thinking about reliability as an engineering discipline, not a dashboard collection. Free online.
- BookTeam Topologies ↗
Most platform problems are team-boundary problems pretending to be tooling problems.
Platform engineering
- BookTeam Topologies ↗
Most platform problems are team-boundary problems pretending to be tooling problems.
- ResourceGoogle SRE Books ↗
Useful for thinking about reliability as an engineering discipline. The Workbook is practical; the original is foundational.
- ResourceKubernetes Documentation ↗
Still the source of truth when blog posts start hallucinating YAML.
- BookPlatform Engineering ↗
Practical guide to building developer-centric infrastructure. Treats platforms as products.
Distributed systems
- BookDesigning Data-Intensive Applications ↗
Makes platform engineering feel less like YAML wrangling and more like systems design.
- ResourceDistributed Systems for Fun and Profit ↗
Short, clear, and still useful. Free online.
- ResourceJepsen Analyses ↗
A healthy reminder that systems lie under pressure. Read before trusting any database vendor marketing.
Security and identity
- SpecOAuth 2.0 RFC 6749 ↗
Painful but necessary if you work near authentication. The spec is clearer than most blog posts about it.
- SpecOpenID Connect Core ↗
Read alongside OAuth. Understand the difference before you build on top of it.
- ResourceSPIFFE and SPIRE Documentation ↗
Useful mental model for workload identity. The concepts transfer even if you use a different implementation.
- ResourceNIST Zero Trust Architecture (SP 800-207) ↗
Useful when kept concrete. Skip the vendor versions — this is the authoritative source.
AI engineering
AI gets interesting when you stop asking "can the model call a tool?" and start asking "what is it allowed to touch?"
- ResourceAnthropic Documentation ↗
Tool use, prompt caching, and the model card. Best provider docs for understanding what the model actually does.
- ResourceOpenAI Evals ↗
Useful starting point for evaluation patterns. The concept matters more than the framework.
- ResourceBuilding LLM-powered applications ↗
Grounded survey of patterns that actually hold in production. No hype.
For Staff+ engineers
- BookStaff Engineer ↗
Interviews with real Staff engineers. Demystifies archetypes and career paths.
- BookThe Staff Engineer's Path ↗
More practical than theoretical. Good for navigating influence without authority.
- BookAn Elegant Puzzle ↗
Systems thinking applied to engineering organisations. Useful whether you manage or not.
- BookThe Engineering Executive's Primer ↗
A level up from Staff. Worth reading before the conversation about scope happens.
The Staff+ job is not "senior engineer, but with more meetings." It is technical judgement at organisational scale.
Things I don't recommend blindly
- ReferenceAdopting Kubernetes because everyone else did
It's the right choice for multi-tenant platforms with many teams. It's expensive overhead for a small product with one team.
- ReferenceBuilding an internal developer platform before understanding the teams
Platform engineering is a service business. Build for the teams you have, not the teams you imagine.
- ReferenceAdding AI agents to workflows without evals and access boundaries
A tool-using agent with no guardrails is a bet that nothing will go wrong.
- ReferenceTreating observability as dashboards
Dashboards are for showing executives that something is fine. Observability is for debugging when it isn't.
- ReferenceCopying FAANG engineering practices
Useful to understand. Dangerous to copy without FAANG-shaped problems, FAANG-sized teams, and FAANG-level investment.