For Junior, Senior & Staff Roles
A complete preparation guide for junior, senior, and staff network engineering roles, built to get you interview-ready in two to three days of focused reading.
Available now on Amazon · by Shane Goodwin
This is not a certification textbook. It assumes you've already worked with networks and need a structured, honest refresher that goes deep exactly where interviews go deep: data center fabrics, BGP, overlays, Kubernetes networking, and troubleshooting under pressure.
2–3 questions per chapter. Fundamentals, definitions, and simple scenarios.
3–4 questions per chapter. Design reasoning, tradeoffs, and hands-on troubleshooting.
4–5 questions per chapter. Fleet-scale thinking, failure domains, organizational judgment.
Where a topic like BGP spans multiple chapters, every chapter attacks it from a different angle, and no question repeats across the book.
The book is written to be read cover-to-cover, but three suggested paths let you go deep exactly where a given loop will push you.
Chapters 1–5 in full, Chapters 6 and 15 at survey level, then Chapter 16. Keep the quick-reference tables and glossary open on interview day.
Skim Chapters 1–5. Go deep on Chapters 6–8, 10–11, and 14–15, then Chapter 16. Appendix B becomes your final-hours review sheet.
Skim Chapters 1–5. Go deep on Chapters 6–9 and 12–15. Spend real time with Chapter 16's worked designs, tradeoff framing, and level calibration.
Sixteen chapters, each ending in leveled interview Q&A.
Three questions, three levels, all from the same book: Chapters 1, 6, and 9.
"Walk me through what happens, at the network level, when you load a web page from your laptop."
"The browser resolves the name first: the stub resolver checks its cache, then asks the configured recursive resolver over UDP 53, which walks root, TLD, and authoritative servers if it doesn't have the answer cached. With an IP in hand, my laptop checks whether the destination is on its own subnet. It isn't, so it forwards to its default gateway…"
"Walk me through a three-tier Clos fabric with 4 spines and fabric switches feeding TORs. Describe the design and its failure domains."
"Three tiers with strict adjacency: TORs at the rack, a fabric tier forming pods, spines on top… Failure domains, quantified: one spine down costs 25 percent of inter-pod bandwidth and zero reachability… And I'd flag the correlated domains that don't appear in the diagram: spines sharing a power feed or fabric switches sharing an optics batch fail together, so physical diversity is part of the failure-domain design, not an afterthought."
"Design the network for a 1,024-GPU training cluster: topology, provisioning, congestion regime, and the constraints that shape it."
"Topology: rail-optimized, 8 rails, each rail a leaf tier serving one NIC rank across all 128 nodes, with a spine layer interconnecting rails… Transport and congestion: RoCEv2 with DCQCN, ECN thresholds tuned per the switch generation's buffer architecture, PFC as backstop with aggressive alarming… The constraints that actually shape it: physics and money in the optics…"
Available now on Amazon in Kindle and paperback.