Levels & tracks — what the bar is, and how the round changes
A system-design round is not one test with one bar. A Staff Engineer and an Engineering Manager can draw the same diagram and get opposite verdicts, because the panel is asking a different question about each: can this person own the hardest technical decisions? versus can this person own a system, its operations and the team that runs it? Senior managers and directors face a third question: can this person set direction for several systems and the organisation around them?
design-room therefore calibrates on two axes: the track (individual contributor or manager)
and the level within it. Pick both at session start, or set them once in profile.md.
1 · The six levels
| Code | Level | Track | Typical titles | The question the panel is asking |
|---|---|---|---|---|
senior |
L5 / E5 | IC | Senior Engineer | Can they design a well-specified system cleanly and go deep when probed? |
staff |
L6 / E6 | IC | Staff Engineer | Can they frame a vague, large problem and make the hard technical calls — and say why? |
principal |
L7+ / E7+ | IC | Senior Staff · Principal | Can they set technical direction across many teams and years, and be right about the bets? |
em |
L6 / M1 | Manager | Engineering Manager | Can they own a system end to end — design, operations, cost — and deliver it through a team? |
senior-em |
L7 / M2 | Manager | Senior EM · Manager of Managers | Can they run several systems and teams, sequence multi-quarter work, and manage risk across them? |
director |
L8 / D1 | Manager | Director · Head of Engineering | Can they set an org’s technical and investment strategy, and make the portfolio trade-offs? |
Default: em. Override per session (Mock me as staff) or in profile.md.
Does the IC/manager split make sense for a design round? Yes — and getting it wrong is a common way to fail. Staff candidates who answer like managers (“I’d have the team load-test it”) without going two levels deep read as under-levelled technically. EM candidates who disappear into B-tree internals for twenty minutes and never mention on-call or cost read as senior ICs who want the title. Directors who design boxes instead of investment decisions read as EMs. The interviewer will tell you which way you drifted.
2 · Dimension weights
All nine dimensions are always scored 1–4 against the target level’s bar. Weights change how much each counts toward the weighted score in the report. H = 1.5 · N = 1.0 · L = 0.5.
| # | Dimension | senior | staff | principal | em | senior-em | director |
|---|---|---|---|---|---|---|---|
| 1 | Requirements & scoping | N | N | H | N | H | H |
| 2 | Estimation | N | N | N | N | N | L |
| 3 | High-level design | H | N | L | N | L | L |
| 4 | Technical depth | H | H | N | N | L | L |
| 5 | Trade-off reasoning | N | H | H | N | N | N |
| 6 | Operations, reliability & cost | L | N | N | H | H | H |
| 7 | Communication & collaboration | N | N | N | H | N | N |
| 8 | Handling ambiguity & pushback | N | H | H | N | H | H |
| 9 | Leadership signal | L | N | H | H | H | H |
Weighted score = Σ(weight × score) ÷ Σ(weight), over scored dimensions only. An unscored dimension never counts as a 2.
2.1 · Dimension 9 means different things on the two tracks
| Track | “Leadership signal” listens for |
|---|---|
| IC | Technical leadership. Sets direction others can follow; designs for the team that will build it; names the decision that needs an RFC and who must agree; anticipates the migration path; mentors through the design (“I’d want the on-call engineer to be able to answer X from the dashboard”). At principal: multi-year bets, platform leverage, and which problems the org should not solve. |
| Manager | Organisational and delivery leadership. MVP-first; sequences the riskiest work first; splits the system along team seams; names staffing, on-call load and the operational cost of the design; zooms out to business impact. At director: investment and portfolio trade-offs, org design, build-vs-buy, risk appetite. |
3 · What changes in the room, by level
The interviewer changes manner and emphasis, never the rubric.
| senior | staff | principal | em | senior-em | director | |
|---|---|---|---|---|---|---|
| Prompt | as written, clarifications answered fully | under-specified | deliberately vague; “the business wants…” | as written | wider: two systems or a migration | a business problem, not a system |
| Who drives deep dives | interviewer picks | candidate is expected to pick the risky component | candidate picks and justifies why it’s the bet | interviewer picks, includes ops | candidate picks, includes sequencing | interviewer steers to org/portfolio within ~15 min |
| Depth expected | two levels on one component | two to three levels on two components; mechanisms, not brands | one level everywhere, three where the bet is | two levels on one component | one level, then trade-offs | box-level is enough; decisions are what count |
| Typical curveball | a component fails | requirements change mid-design | a constraint invalidates the strategy | an incident or a cost spike | a team or a dependency disappears | budget cut 30%, or a regulator |
| Wrap emphasis | bottlenecks | trade-offs you’d revisit | what you’d stop the org doing | SLOs, on-call, cost | sequencing and risk register | investment, org shape, metrics |
3.1 · Archetype fit
| Level | Where loops usually draw from | Most revealing in this bank |
|---|---|---|
senior |
A–G, I, J (build rounds) | SD-02, SD-20, SD-25, SD-38, SD-47 |
staff |
A–G, I, J at depth, K | SD-15, SD-19, SD-28, SD-66, SD-71, SD-84, SD-87 |
principal |
C, D at strategy depth, K | SD-27, SD-33, SD-83, SD-85, SD-86, SD-88 |
em |
all build archetypes + H | SD-08, SD-44, SD-55, SD-57, SD-58, SD-59 |
senior-em |
H, K strategy, F/D at scale | SD-56, SD-59, SD-60, SD-90, SD-92 |
director |
H, K strategy rounds | SD-60, SD-62, SD-89, SD-90, SD-91, SD-92 |
Every question’s level fit is in the Levels column of Question-Bank.md.
4 · From weighted score to outcome
| Weighted score | Outcome at the target level |
|---|---|
| ≥ 3.25 | Strong hire |
| 2.75 – 3.24 | Hire |
| 2.25 – 2.74 | Lean no / mixed — usually one dimension sank it; the report names it |
| < 2.25 | No hire |
Two overrides, both applied by the interviewer and stated in the report:
- A 1 on a High-weight dimension caps the outcome at “lean no”, whatever the average. A staff candidate with no depth, or an EM with no operational story, does not get hired on charm.
- Apparent level. The report also says where the round actually landed — e.g. “staff-level depth, senior-level framing”. If that differs from the target, it is the most useful line on the page.
5 · Choosing your level honestly
Practise one level above where you are confident. The round is calibrated to the target, so a 3
at staff is worth more than a 4 at senior, and the gap is shown to you in dimension terms rather
than as a vague “not quite”. When you consistently score ≥3 at the target, move up for a week and see
which dimension breaks first.