Keentune

Engineering Leadership curriculum

21 chapters
·
132 concepts
·
free
Everything the adaptive question bank can teach and test in Engineering Leadership, from foundations through advanced practice. Work through it in order, or start practising and let the questions find your level.
New here? Read the Engineering Leadership guide
A free 17-minute primer — the mental model, the mistakes beginners make, and what to practise first.
A. Stepping into leadership
a lead is measured by what the team ships, never by their own commit count
one meeting costs a maker the half-day around it, not the hour inside it
name the changed relationship once, explicitly, rather than hoping it settles itself
taking the critical-path task yourself makes you the bottleneck and starves the team of growth
DORA measures vision, intellectual stimulation, inspirational communication, supportive leadership, personal recognition
good leadership lifts delivery only by enabling technical and process capabilities, not on its own
B. Delegation and decision rights
hand over the definition of done and the constraints; prescribing the path is supervision
the same task can be delegated as "do it and report", "recommend and I decide", or "decide and act"
how directive to be depends on experience with *this* task, not on overall seniority
the test of a delegation is who owns the next step when the conversation ends
work done 80% as well as you would have done it is a success, not a defect to correct
one-way doors deserve a heavyweight process; two-way doors should be decided fast and locally
C. 1:1s, coaching and motivation
the report sets the agenda; the lead's questions fill the gaps
a 1:1 that becomes a status readout burns the only private channel you have
goal, reality, options, will: the conversation ends with the person choosing an action
a skip-level samples how a team is being led; using it to redirect work undercuts the manager
for complex work, intrinsic motivators outperform contingent rewards
pay and conditions cause dissatisfaction when bad but do not motivate when good
D. Feedback that changes behavior
situation, behavior, impact: when it happened, what was observed, what it caused
"you interrupted twice" is actionable; "you're dismissive" is a verdict
the giver reports the impact they saw; the receiver owns the intent, and the two rarely match
burying criticism between praise makes the praise suspect and the criticism invisible
care personally and challenge directly are independent axes, and both are required
caring without challenging is the most common failure, and it feels like kindness
challenging without caring is not honesty; it closes the channel it needed
E. Hard conversations and performance management
nothing in a review should be the first time the person hears it
open with the conclusion; a softened lede is heard as reassurance
diagnose skill, motivation or system first, because the three call for different responses
a verbal warning the person can reinterpret later is not a warning
run it honestly as one of the two; a plan designed to fail is a trust-destroying formality
the bill is paid by the team's standards and by your strongest people
F. Psychological safety and the culture that produces it
a shared belief the team is safe for interpersonal risk, not a belief that the team is comfortable
high safety with high standards is the learning zone; safety alone is the comfort zone
Edmondson's hospital study found better-led units reported *more* errors, inverting the original hypothesis
safety is set by what happens the first time someone brings the senior person a problem
pathological cultures hoard information, bureaucratic ones defend rules, generative ones serve the mission
Westrum-generative culture is a measured predictor of delivery and organizational outcomes
G. Team dynamics
forming, storming, norming, performing; storming is a stage to work through, not a failure
absence of trust, fear of conflict, lack of commitment, avoidance of accountability, inattention to results
people commit to decisions they could argue against, not to decisions announced at them
dissent belongs before the decision; relitigating it afterwards is the failure mode
individual effort drops as group size grows unless each contribution stays visible
H. Hiring, interviewing and onboarding
the same questions plus an anchored rubric predicts performance; a free-form chat mostly measures rapport
sampling the actual work predicts strongly; brainteasers and pedigree predict weakly
written scores submitted before the debrief stop a senior voice anchoring the panel
"fit" quietly selects for sameness; state the values required and the perspective missing
an interviewer with no stake in filling the seat, because the hiring manager's urgency lowers the bar
an unresolved substantive doubt is a no; a mis-hire costs more than a missed hire
hire juniors for learning rate, seniors for demonstrated depth
a new hire deploying something small in week one is the ramp signal that matters
I. The four keys and what they measure
deployment frequency, lead time for changes, change failure rate, failed-deployment recovery time
the highest performers are faster *and* more stable, so the trade-off is a symptom of large batches, not a law
lead time for changes runs from code committed to code running in production, not from idea to launch
it is measured because batch size itself is hard to measure directly
the share of deployments that degrade service and need remediation, not the share of tickets that are bugs
restoring service quickly matters more than failing rarely, because failure is certain
they describe a delivery system; attributing them to one engineer is a category error
each key is gameable alone, but throughput and stability push against each other
J. Batch size, queues and flow
a smaller change is easier to review, faster to verify and cheaper to revert
defect rate, rework and elapsed time all scale with how much is in flight at once
a system loaded near capacity has wait times that grow non-linearly
limiting work in progress finishes things sooner; starting more finishes less
order work by value lost per week of waiting, not by size or by who asked loudest
K. Estimation, commitments and Brooks's law
an estimate is a distribution, a commitment is a promise; conflating them manufactures failure
early estimates are uncertain by multiples and narrow only as the work is actually done
the moment the date moves in your head is the moment to say it out loud
of scope, date and quality, scope is the one that can honestly move
adding people to a late project makes it later: ramp-up cost plus new communication paths
the law bites hardest on work that cannot be partitioned, which is the honest boundary of the claim
L. Meetings and group decisions
decide, discuss or inform; a meeting with no stated type defaults to status
a discussion with no decider produces consensus theatre or no decision at all
invite the smallest set that can reach the outcome; everyone else gets the notes
attention drifts to the trivial item everyone feels qualified to debate
imagining the failure before committing surfaces risks a status review never will
written rounds and explicit turns surface the quiet read the loudest voice buries
M. Technical strategy, debt and migrations
capture context, options, decision and consequences; the reasoning is the artifact, not the choice
build where the capability differentiates the product; buy the rest
frame tech debt to non-engineers in weeks of future delivery and incident risk, never in aesthetics
a big-bang rewrite discards years of absorbed bug fixes and delivers nothing until the end
migrate by routing traffic incrementally so every step is small and reversible
with no date to delete the old system you have permanently doubled the surface area
DORA finds an external change-approval board slows delivery without improving stability
N. Org design, Conway's law and Team Topologies
a system's structure mirrors the communication structure of the organization that built it
reshape the teams to produce the architecture you want, instead of fighting the org chart in code
teams able to test and deploy without depending on other teams is the strongest measured architectural predictor
stream-aligned, platform, enabling and complicated-subsystem
collaboration, X-as-a-service and facilitating, with collaboration deliberately time-boxed
a team owns what it can hold in its head; extra services do not fit just because headcount grew
report count trades depth of support against layers of hierarchy, and both sides cost something
a reorg resets relationships and context, and that dip belongs on the cost side of the decision
O. Incident command
one person owns coordination and decisions, and does not debug
command, operations and communications are separate so the person fixing is not the person briefing
under-declaring costs more than a false alarm; escalation is cheap and delay is not
one shared timestamped document is what makes handover and the later postmortem possible
restore service first; root cause is a postmortem activity, not an outage activity
roll back unless rollback is impossible or riskier than the forward fix
executives wanting updates go to the comms lead, never to the person typing
P. Blameless postmortems
ask what made the mistake easy, not who made it
"human error" ends the investigation exactly where it should begin
the signal looks obvious afterwards; judge the decision on what was knowable at the time
unowned follow-ups make the document the deliverable instead of the fix
deciding beforehand which events get a postmortem keeps the choice from becoming political
a five-whys line hides the several conditions that all had to align
Q. On-call, error budgets and toil
a rotation needs enough people that nobody is paged while exhausted
an alert no human can act on should be deleted, not tolerated
an SLO below 100% creates a budget of unreliability that funds change
what happens when the budget is spent is decided before it is spent
manual, repetitive, automatable, tactical work that scales linearly with the service
capping ops work protects the engineering time that would remove it
a service only one person can rescue is a single point of failure with a pulse
R. Stakeholders and upward communication
a project green until it is suddenly red has a reporting problem, not a schedule problem
escalate with the choices and the one you would take, not with the problem alone
stakeholders buy dates, risk and money; the architecture is your reasoning, not their decision
escalating jointly with the peer lead unblocks; escalating around them costs the relationship you still need
a dashboard anyone can read beats a status report only you can produce
the honest answer names what would have to move, rather than a flat refusal or a silent yes
S. Careers, ladders and staff-plus
a ladder is defined by the scope and ambiguity a person handles, not by tenure
promotion confirms someone is already operating at the level; it is not a bet on potential
cross-manager calibration exists so one manager's "exceeds" means what another's does
tech lead, architect, solver and right hand are four different jobs under one title
a mentor gives advice; a sponsor spends their own credibility on your behalf in a room you are not in
the coordination that makes a team function is often uncounted at promotion time and unevenly assigned
T. Measuring engineers without breaking them
a measure that becomes a target stops being a good measure
commits, lines and story points count activity, are trivially gamed, and ignore the system
features shipped is output; the change in user or business behavior is the outcome
satisfaction, performance, activity, communication, efficiency; never report one dimension alone
the original variance studies confounded environment with person and never established a per-person multiplier
U. Remote and async practice
a decision that exists only in a call did not happen for anyone who was not on it
DORA finds good internal documentation amplifies the effect of every other capability
with one person remote, either everyone dials in or the remote person is a spectator
a team spread past a few overlapping hours must change its process, not work later
visibility in an office gets mistaken for contribution when ratings are written
Keentune is not affiliated with or endorsed by the organizations whose documentation informs these maps.
All about Engineering Leadership practice
Also on your phone
All exam, test, and product names and trademarks are the property of their respective owners and are used here for identification and reference only. Keentune is independent study practice — not affiliated with, authorized, or endorsed by any of these organizations.
© 2026 SportaApp LLC