domain-ramp-upGet competent in an unfamiliar domain such as payments, logistics, ad-tech, clinical billing or telecom in about two weeks, in a fixed order: one page of the money, data and decision flow before any vocabulary, a glossary capped at 30 terms each with a real example record, one real transaction trace
exam-prep-planWork backwards from a fixed certification or exam date such as CKA, AWS, CISSP or a university module to a study schedule. Day 0 is a full-length practice exam scored by blueprint weight; weight times miss rate ranks domains and sets hours. Schedule backwards in three blocks (gap closing, mixed prac
exercise-designWrite a programming or operations exercise, lab or kata that teaches one thing and can be done unattended. Built from the end: the working solution first, then failing tests or a check script that gives the learner a clear signal within two minutes, starter code that compiles and runs, and a three-t
learn-new-technologyLearn a framework, database, language or tool well enough for real work in the first week. Order: run the official minimal example unmodified, then break it on purpose and read the error; read concepts before the API reference, naming the three core abstractions and what the tool does not do; read t
literature-scanSurvey the state of the art on a technical question before building. Write the question in one sentence with in-scope and out-of-scope criteria, seed with 3 to 5 papers from a recent survey or leaderboard, snowball backward and forward two hops. Stop when two consecutive rounds of ten candidates add
mentoring-cycleRun a recurring mentoring relationship as the mentor or the mentee, outside the reporting line. The first session sets a contract: two goals at most, a fortnightly 45-minute cadence, confidentiality, and a no-offence way to end it. The mentee owns the agenda and sends it 24 hours ahead; the mentor a
paper-readingRead one research paper efficiently and decide what it means for your work. A five-minute triage reads the abstract's claim, the one results table and the limitations: read, skim or skip. The second pass reads figures and tables before the method. The claim is forced into one sentence, on X, method
paper-reproductionReproduce or port a published method to see whether its claimed result holds on your data. Go or no-go first: reproduce only what you would deploy. If the authors' code exists, run it unchanged and match one reported number within the paper's own seed variance before touching anything; otherwise imp
reading-backlogGet a hopeless pile of saved articles, books, papers and bookmarks under control. Count and date the backlog first; anything older than 90 days goes into a searchable archive with no shame, out of the queue. Each item left must name the task or decision it serves this quarter: no task, no read. Keep
shadow-reverse-shadowHand a recurring duty such as releases, deploys, on-call, a monthly close or a vendor relationship to another person safely, by doing rather than telling. Four dated rungs, each with a written gate of three to five observable actions and a count: watch with narration, assist, do-with-watcher (revers
skill-acquisition-planPlan learning a substantial new skill over weeks or months from a stated goal. Start from a proof artefact someone else can inspect, with a date (a deployed service, a passing practical exam, a talk given); decompose backwards by prerequisite and tag every sub-skill \"know about\" or \"can do\", cou
tacit-knowledge-elicitationExtract what a departing or overloaded expert knows but has never written down, as decision rules rather than a transcript. Asking \"what do you know\" omits the judgement; instead run three recorded 45-minute critical-incident interviews, each built on a real past case, asking what they checked fir
teaching-pairingPair with a junior or newcomer to teach, not to deliver the task through them. The learner drives, always; replace telling with a three-rung ladder of question, hint, then showing, waiting a count of ten before moving up; the learner predicts each run and narrates intent; sessions capped at 90 minut
team-training-curriculumBuild a multi-week internal training path for a team or cohort. Starts from a task list (\"by Friday they can roll back a release alone\"), measures the gap from real work (PRs, incidents, tickets) not a survey, sequences modules by prerequisite and just-in-time use (taught within two weeks of need)
training-effectiveness-checkFind out whether a workshop, course or training programme changed what people do, not how they rated it. Choose the measure before delivery; run the same practical task before and after (a lab, a review of a sample change), not a quiz; name one behaviour indicator seen in real work and read it at 30
workshop-designDesign a hands-on technical workshop of two to eight hours for colleagues or a public audience so it stays a workshop and does not turn into a lecture. Three outcomes written as observable tasks, each verified by a closing check; hands-on time of at least 60 per cent with no lecture segment over ten