Establish as a subject-matter expert (data / AI / ClickHouse)
Direction (decided 2026-07-01)
Don’t be an SME in a topic — be one in a method. Brand: rigorous data-infra investigator. You settle arguments with experiments. The investigation is the product; ClickHouse is the current lab.
This dissolves the 2026-06-30 prerequisite split (“depth first, then positioning”): doing rigorous experiments publicly is the depth-building. The investigation and the positioning are the same activity. No separate prerequisite phase.
Execution
- 10 hrs/week, ship every other Friday
- Week 1: pick one claim, build a clean repro, get the numbers
- Week 2: write it up with receipts and publish
- Topic source: customer-work backlog. Momentum over polish.
Audience
Developers are distribution; architects and heads-of-data/platform are the value target — the person whose job is on the line if the decision is wrong. Filter every topic: whose ass is on the line if this goes wrong?
Attitude
Neutral investigator, not advocate. “Here’s the claim, here’s what actually happened.” Falsifiable beats opinionated. Boring findings still ship.
Publishing
- Internal (ClickHouse): frame it so the company looks good → encouraged, not policed. Use the CH blog where interests align; being known inside reinforces being known outside.
- External: canonical copy on your own site (survives any employer change), then syndicate for reach. Brand as data-infra investigator — not “the ClickHouse guy.”
Success signal
Unsolicited inbound per quarter.
Still open
- Manager vs principal-engineer track — the direction above assumes principal/external-artifact path. If the track tips managerial, this strategy is largely off-strategy; internal scope/influence becomes the lever instead.
- Nature-business vs SME as primary identity bet — are these one thread or two competing threads for the same scarce hours? Which one gets the next 3 years?
- What’s already there? Any existing writing, talks, OSS to build on rather than start from zero.
Background
Original question
How does Denys establish himself as a recognized SME in data/AI/ClickHouse — market-relevant and market-attractive independent of any employer?
2026-06-30 reframe
The original question presupposed an existing depth substrate. Denys flagged the presupposition was wrong: ClickHouse consulting engagements are short and slice-scoped — they don’t accumulate into a cross-cutting view. The question split into: (1) build depth first; (2) position only after. Forcing #2 without #1 = misery lane (shallow content, uncredible). Named structural risk: if the seat can’t produce SME-depth, this is a seat-shape problem, not a content-strategy problem — connecting to the manager-vs-principal decision and the 5–10-year role question.
Depth-building paths explored: deeper inside one engagement (A), internal CH engineering depth (B), publicly commit to a lane (C). Positioning bets considered: consultant cross-cutting view (1), adjacent-layer positioning (2), artifacts > articles (3).
2026-07-01: method-brand direction chose effectively C + 2 + 3, and dissolved the prerequisite split — the investigations are both the depth and the output.
Decision log
- 2026-06-29 — Question raised. Motivation: ClickHouse role is great but creates identity dependency that needs hedging via public reputation. No channels, no success signal, no time budget. Filed as big question: multi-year, identity-shaping, competes with other commitments.
- 2026-06-30 — Re-surfaced. Constraints named: engagement must not be misery, ClickHouse-content lane is already saturated internally, single-product seat removes the variety that drove the consultancy-era content engine. Three positioning bets added (see §Background). Career-track bifurcation named as load-bearing dimension.
- 2026-06-30 (later) — Prerequisite problem surfaced: no existing depth substrate. Engagements are short and scoped — no cumulative pattern view accumulating. Reframe: depth-building must precede positioning. Three depth-building paths explored (see §Background). Structural risk named explicitly.
- 2026-07-01 — Direction locked. Method-brand: rigorous data-infra investigator. Dissolves the prerequisite split. Execution: 10 hrs/week, biweekly cadence, topics from customer backlog. Audience: devs distribute, architects/heads-of-platform are the value target. Attitude: neutral investigator, falsifiable. Internal: CH blog where aligned. External: own site canonical, syndicate for reach. Success signal: unsolicited inbound per quarter.