Career development

Umbrella note for deliberate career growth — the technical and human skills compound differently and deserve separate shelves: Hard skills (the craft) and Soft skills (the leverage).

Books

• swyx — The Coding Career Handbook (learn in public) — learninpublic.org: careers as compounding loops; ship, write, and teach in the open so opportunity finds you; find your “spike” of differentiated depth rather than spreading evenly. (capture pending — this entry seeded from memory of the book’s thesis; verify + capture before quoting further)

Job search mechanics (the constraint-navigation lane)

• The 90-day notice-period playbook (engineerturnedrecruiter.com guide, captured → clippings/engineerturnedrecruiter-com-how-to-land-a-job-if-you-have-a-90-day-notice-period.md; pointer tweet) — notice period as a screening criterion, not a verdict: strong profiles get filtered before recruiter contact, so manage the number, not just the skills. The sequence: test yourself before resigning (mock interviews + a small application batch as reality check — “being good at your current job and being good at interviews are two different things”), use the first 30–40 notice days as the prep window, apply aggressively once notice drops under 30 days (the filter loosens), don’t rely on applications alone (referrals), keep the remaining-notice field current, treat every interview as feedback — and if interviews aren’t coming, check the profile before blaming the constraint. The structural read underneath: companies hire against business urgency, and mutual distrust (“both sides optimise for themselves”) makes joinability a tiebreaker between similar candidates. Resigning without an offer gets a no-universal-answer treatment — financial runway × tested skills × market reality. Market-specific to India’s 90-day norm but the pattern (a screening constraint you demonstrate your way under) generalizes.

Growing

• (open: L3→L4→L5 progression notes, promotion packets, the COT L3 standards angle)
• (open: mentorship/sponsorship distinction; conference talks as career lever)

Seniority as scar tissue

• Senior mindset — the done-not-read checklist (rollbacks, postmortems, zero-downtime migrations, idempotency keys, alerts that fired, EXPLAIN, load tests, mentoring through first oncall, saying no); incident-shaped + design-for-failure + force-multiplier triad; interview question = “tell the story behind each item.”

• Sysops interview questions — the sysadmin/sysops interview canon (TecMint 20 networking Qs captured + the L3 lanes: storage, perf, systemd, incident walks); Q20-style debugging walks beat definitions in the LLM era.

The manager-track counterweight, from the Polish feeds (2026-10-08)

FinansowyUmysl (125❤/12.4k views, PL; quoting klatkowaelita): the manager role itemized as “tłumaczenie się z nierealnych celów… rozliczanie ludzi z nierealnych celów… odmawianie ludziom podwyżek… wybieranie osoby do zwolnienia… szukanie nowej osoby za niższą stawkę niż zwolniony i ściemnianie mu jak tu jest super… branie odpowiedzialności za wynik, który w ogóle od Ciebie nie zależy — a to wszystko za minimalnie większą pensję niż osoba pod Tobą.” His yearly ritual: re-evaluating the manager path, “minusy mocno przeważają nad plusami.” The quoted frame: the modern manager role is “idealne stanowisko dla kogoś z całkowitym deficytem empatii — empatia to w tym systemie zbędny koszt.” For this note’s ladder lanes: the strongest IC-staying argument in the vault — the anti-Schmidt datapoint, where the glue-people counterweight (Senior mindset) meets its floor: when the org treats empathy as cost, the glue role is extraction, not leverage. Kin: Senior Engineer Mindset (book) (Swizec’s IC-path thesis), Senior mindset (Schmidt glue-people as the mirror image).

Karwatka: the agency that compounds — assets that survive the invoice (2026-10-08)

tomik99 (Tomasz Karwatka; 39❤/161k views) → his Substack (both captured): the co-founder of Divante and Callstack (Vue Storefront, Open Loyalty) on turning services work into owned assets — the Polish-ecosystem counterpart to this note’s career-compounding lanes. The core question: revenue pays, the case study sells, the team learns — “but how much of that learning makes the next project easier, safer or more profitable?” Load-bearing moves: experience needs a place to live (“if every important lesson lives in someone’s memory, the business has a fragile way of keeping that asset available” — reusable assets = diagnostic processes, integration patterns, test lists, delivery checklists, small domain components); AI raises the value of explicitness (“a person or an agent can work from clear examples, documented decisions and acceptance criteria” — skills as the packaging); repeated demand beats leftover code (productizing a client pull-out is harder than it sounds — “the original solution fits one client’s assumptions”; first ask whether several independent clients have the same underlying problem); the narrow business case (review last ten projects → one costly repeated activity → improve within a working month → the test: can a second person use it successfully? — “a component that works only when its author operates it has not yet reduced dependence on that person”); and the maintenance-honesty: abandoned internal tools create effort too. This is the business-owner’s version of the same compounding thesis as swyx’s learn-in-public — and the “second person can use it” test is the agency-scale restatement of skills’ certification seam (agent-usable = no single-author dependency).