Skip to content

Know what you're missing before your Senior or Staff engineering interview

For backend and fullstack engineers. A readiness check measures the competencies our Senior and Staff targets cover and tells you which are Ready, Borderline or a Gap. Then you practise exactly those, round by round. Both come with a pass.

Free, with no account: the demo, with 40 questions and 2 coding exercises. Everything else needs a pass: the readiness check, the question bank, quizzes, packs, flashcards and the other 22 exercises.

Already have an account? Log in

An example readiness report, with made-up answers

Readiness for Senior · backend
System design & trade-offs: Ready, 3 of 3 correctDistributed systems: Borderline, 2 of 3 correctData modelling & storage: Ready, 2 of 3 correctReliability & failure modes: Borderline, 2 of 3 correctSecurity: Ready, 2 of 3 correctAPI design & contracts: Ready, 3 of 3 correctConcurrency: Gap, 1 of 3 correctDebugging and incident reasoning: Borderline, 1 of 3 correctSystem designDistributedData & storageReliabilitySecurityAPI designConcurrencyDebugging
ReadyBorderlineGapReady bar for Senior

Highest-risk gap: Concurrency

Every answer cites the source it was checked against, so you can check the reasoning instead of memorising someone's answer key.

A question from the free demo

A tenant namespace runs a Deployment of twelve replicas whose pod template runs as root. On Monday the platform team adds the label pod-security.kubernetes.io/enforce: restricted to the namespace; the label is accepted and all twelve pods keep serving all week. On Saturday a node-pool upgrade drains the nodes one at a time, and by the end the Deployment has no running pods, though nobody touched it. Why did the Deployment fail on Saturday rather than on Monday, when the label was added, and what should rolling out such a label have included?

Answer: Pod Security admission judges requests: enforce rejects a pod when it is created, so the twelve already running were never re-evaluated. The drain evicted them, and every replacement the ReplicaSet created was a new pod the restricted profile refused, so capacity fell node by node. Rolling out enforce needs warn and audit first, which also flag the workload, and templates fixed before enforce.

Read the source

A coding exercise from the free demo

The config that panics on write

LoadDefaults is supposed to return a config map with defaults applied, and callers then add their own keys. It compiles, it reads fine, and it panics the moment anyone writes to it. Fix it so the returned map is usable.

func LoadDefaults(overrides map[string]string) map[string]string

Examples

  • LoadDefaults(nil) → {"region":"eu-central-1","retries":"3"}
  • LoadDefaults(map[string]string{"retries":"5"}) → retries overridden to "5"
  • the returned map must accept new keys without panicking

Constraints

  • Defaults are region=eu-central-1 and retries=3
  • Any key in overrides wins over the defaults, including keys that are not defaults
  • Do not modify the overrides map the caller passed in

Write it, run the tests and see which cases fail, in your browser and with no account. The Go interpreter runs on your machine and your code is never uploaded.

Open this exercise

Check the source behind every answer

2058 questions across 18 categories: AWS, System Design, Go, AI Agents, Kubernetes, Databases & SQL, Python, Algorithmic Foundations, Messaging & Event Streaming, Google Cloud, Networking & APIs, Behavioral, Observability, Microsoft Azure, Frontend & React, Infrastructure as Code, TypeScript & Node, Design Patterns. Every explanation cites a checkable source (documentation, specifications, books), not exam dumps.

A readiness check per competency

Part of every pass. Each competency is judged against your target level as Ready, Borderline or Gap, and the highest-risk one is named. Then you see progress per competency as you close the gaps.

The example report above, axis by axis

  • ReadySystem design & trade-offs3 of 3
  • BorderlineDistributed systems2 of 3
  • ReadyData modelling & storage2 of 3
  • BorderlineReliability & failure modes2 of 3
  • ReadySecurity2 of 3
  • ReadyAPI design & contracts3 of 3
  • GapConcurrency1 of 3
  • BorderlineDebugging and incident reasoning1 of 3

A hard question counts for more: 2 of 3 is Ready on data modelling & storage, where the hard one was right, and Borderline on distributed systems, where it was not.

  • Go exercises you run in the browser

    24 exercises: most hand you code with a bug to find, a few start from an empty function. Each is checked against real tests, and nothing you type is uploaded.

  • Timed quizzes and repeatable packs

    103 fixed packs you can take again, so you can tell improvement from a lucky run.

  • Flashcards with spaced repetition

    What you keep getting wrong comes back sooner, and what you know comes back later.

  • A study plan, if you have weeks

    Work through it item by item, or go straight to the quizzes if your interview is days away.

  • Questions at Senior and Staff level

    Trade-offs and failure modes, not definitions. A Staff check adds the judgment and leadership axes of the Staff target.

What are you interviewing for?

Each profile says what we check for that target, round by round, and what is here to practise for it, generated from the same data the readiness check reads.

Interview in a couple of weeks: take the check, drill the gaps. Preparing over months: the study plan lists what to cover, led by the gaps your check found, and spaced repetition schedules each flashcard's next review. Same product, one-time pass.