Developer Relations interview prep.

AI DevRel sits at intersection of (a) AI-fluent technical credibility (must ship working AI demos, write robust prompts, design evals, debug refusals - not call an API), (b) AI-builder community (app devs / ML engineers / prompt engineers - each a different mental model), (c) content + advocacy...

What interviewers look for

  • Is the candidate AI-fluent + a credible engineer - can they ship a working demo, write a robust prompt, design an eval, handle a refusal live - or are they a marketer who reads release notes?
  • Do they tie content + community + advocacy to a measurable developer outcome on the AI platform (activation, TTFHW on the quickstart, eval-quickstart completion, capability adoption) - not vanity (views, demo applause, conference attendance)?
  • Can they design a docs + content strategy that survives quarterly model updates (Diataxis with AI overlay; quickstart + eval cookbook + agent recipes + safety playbook; versioned content)?
  • Do they build community deliberately around the AI-builder personas - one home, code of conduct, content cadence, cookbook contributors, champion recognition - rather than 'we will start a Discord'?
  • Can they run the two-way bridge - translate developer friction into capability + safety + DX feedback as structured evidence to model + product teams, and bring product / model changes back credibly?
  • Are they strategic + measurable - 90-day plan, AI-specific measurement framework, budget defence when token cost is the dominant unit economic - or just creating content?

Behavioural questions to expect

  1. Walk me through your CV.

    What it tests: Story coherence + AI-fluent technical credibility + DevRel impact. Teams want evidence the candidate is a credible engineer who can ship working AI demos + write robust prompts + design evals (so AI builders respect them), with progressively-scoped DevRel work that moved measurable AI-platform metrics - not pure marketing, not 'used an LLM at my last job'.

  2. Tell me about a piece of AI content, an AI-builder event, or a community moment you are most proud of.

    What it tests: Real impact + measurement + AI-fluency under scrutiny. Tests whether the candidate frames a piece of work as audience -> capability hypothesis -> format -> distribution -> measurable outcome - not 'I shot a cool LLM demo video'.

  3. Tell me about a weakness, a failure, or feedback you have received and worked on.

    What it tests: Self-awareness + AI DevRel discipline. Cross-role canonical. Fake weaknesses downgrade immediately. AI DevRel mistakes (shipped a demo that broke on the next model release, taught a prompt pattern with no eval, missed a refusal failure mode in a public demo, brought capability feedback to the model team as opinion not evidence) carry real adoption + brand + safety cost.

  4. Why AI DevRel - and why this vs ML engineering, AI PM, or generic devtools DevRel?

    What it tests: Authentic fit for the AI-specific multi-disciplinary DevRel seat: AI-fluent technical + creative + community + strategic, with capability-shifting + safety-aware overlays. Tests whether the candidate WANTS the AI two-way bridge + the public-facing role + the measurement discipline + the quarterly model-update cadence.

  5. Which AI-builder audience or AI product surface would you want to focus on, and why?

    What it tests: Genuine fit + grasp of how AI DevRel sub-disciplines differ. Tests whether the candidate has a reasoned preference (content-heavy / community-heavy / advocacy-heavy / Head-of-DevRel; foundation-model API vs agent platform vs vertical AI vs horizontal assistant vs AI dev tooling; application developers vs ML engineers vs prompt engineers / AI builders).

  6. Why this firm?

    What it tests: Whether the candidate has done the homework. Bar: firm-specific evidence from the product, model strategy, developer audience, community, content - not generic 'great AI company'.

  7. How would you describe this firm's product + AI developer community in your own words?

    What it tests: Whether the candidate has internalized HOW the firm wins with AI builders - product surface, model strategy, audience, community health, content + docs quality, safety posture - not just that it 'has an API'. Tests whether they have used the product, read the model cards, scanned the cookbook.

  8. How does DevRel actually create value at an AI/ML platform firm?

    What it tests: Whether the candidate understands AI DevRel economics: faster developer activation on a capability (compounds capability adoption + token revenue), community-led support (reduces ticket volume in a domain where developers are confused fast), capability + safety feedback (shapes next model + product release), brand + AI-builder recruiting + cookbook compound credibility.

Technical concepts to master

Capability-aware AI content + docs strategy

Diataxis with AI overlay
Four content types extended: Tutorials (learning, e.g. quickstart), How-To (task, e.g. 'how to build RAG'), Reference (information, e.g. API + model card), Explanation (understanding, e.g. capability deep-dive). Plus AI-specific assets: eval cookbook, agent recipes, safety playbook, model-update migration guides.
Quickstart + TTFHW + eval-quickstart
Quickstart = signup to first working call. Eval-quickstart = a parallel 'in 15 min you have a working eval for your task'. Both are highest-leverage assets; success measured by TTFHW + eval-quickstart completion rate.
Cookbook + agent recipes + sample-code discipline
Working cookbook of agents, RAG patterns, tool-use chains, capability demos; the bridge from quickstart to real-world use; AI builders learn by remixing recipes more than by reading reference.
Versioning for model updates + safety playbook
Every piece of content tagged with model version + dated; safety playbook explains refusals + safe-by-default prompts + responsible deployment patterns; content reviewed every model release cycle.

AI-builder community + cookbook contributors + champions

One home + AI-aware Code of Conduct
Pick ONE primary AI-builder channel (Discord / Discourse / code-host Discussions); enforce a clear Code of Conduct from day 1 with AI-specific norms (no jailbreak prompts shared, no scraped output, responsible disclosure of model failures); moderate consistently.
Content + Q&A cadence keyed to model updates
Weekly content rhythm (capability spotlight + cookbook PR review + model-update notes + how-to) + visible Q&A response (target answer-question rate >80% within 24h); spike cadence around model releases.
Cookbook contributors + champions / MVPs program
Formal recognition of top cookbook contributors + AI-builder community members (early model access, ambassador title, conference invite, paid speaking opportunities, sticker / swag).
AI-builder community health metrics
Active contributors (week-over-week), answered-question rate, response-time, cookbook PR throughput, eval-quickstart completion, capability adoption signal, NPS / DSAT from a periodic AI-builder survey.

AI advocacy + the capability / safety feedback loop

AI-builder feedback aggregation
Systematic harvesting: community Q&A clustering, support ticket categorisation, cookbook PR comments, quickstart abandonment, docs analytics (drop-offs), AI-builder satisfaction surveys, social signal (X, Hacker News, AI subreddits).
Structured capability / safety brief
Friction surfaced as: problem statement + evidence (quotes + numbers + sessions + repro eval) + proposed fix (model behaviour change vs prompt pattern in docs vs refusal copy update vs SDK affordance) + size of win + recommended owner (model / product / safety / docs).
Partnership through ship - including correct safety boundaries
Work with model / product / safety from RFC -> design -> ship; AI DevRel brings AI-builder context throughout; if a safety boundary is CORRECT, partner on safe-pattern communication + workaround for developers instead of pushing to remove the boundary.
Closing the loop with AI builders
Public update when feedback ships: changelog post, blog, video, cookbook update, model card update, community announcement; explicit 'you said, we shipped' framing.

AI DevRel measurement + the 90-day plan

The 90-day plan for an AI DevRel seat
Days 1-30: listen + assess (ship the quickstart + eval cookbook yourself, talk to AI builders across personas, audit channels). Days 30-60: pick + seed (one primary channel, content cadence keyed to model updates, eval-quickstart push, initial wins). Days 60-90: scale + recognize (cookbook contributors / champions, UGC, scaled content, model-update playbook) + first measurement readout.
Leading vs lagging metrics with AI overlay
Leading: TTFHW on quickstart, eval-quickstart completion, cookbook PR throughput, content engagement, response-time. Lagging: activation rate, MAU developers, capability adoption, token usage growth on touched cohort, conversion to paid, NPS.
Attribution + instrumentation in an AI platform
UTM-tagging content + tracking signup -> activation -> capability adoption by source; analytics on docs + cookbook (search, drop-off, copy events); funnel from content view -> signup -> first call -> first eval -> capability adoption -> token usage.
Budget defence with token-cost reality
AI DevRel ROI: activation lift x capability adoption x token revenue per active developer x ARR; support deflection $ saved; recruiting pipeline value; cost-aware patterns saving margin; brand (acknowledged as hard-to-quantify).

Practical drills

  • Walk me through a piece of AI content or advocacy work you produced - the AI-builder audience, the format, the distribution, and the AI-platform metric it moved.
  • Walk me through how you would build (or grow) this firm's AI-builder community in your first 90 days. Sketch the 30 / 60 / 90 milestones + the measurement.
  • Community signal indicates this firm's AI builders are hitting a friction - tool-use chains break past 3 steps, OR the model is refusing a legitimate use case, OR RAG retrieval feels unreliable. Walk me through how you would diagnose + drive the fix.

Smart-question anchors

  • DevRel team + scope - shape, AI surface coverage, what this role would own in 6-12 months
  • Existing AI-builder community + content - current channels, cookbook health, model-launch moves
  • Product + model strategy + developer audience - foundation-model strategy, AI-builder personas, activation path
  • Measurement - AI DevRel KPIs, capability adoption tracking, CFO / leadership relationship under token-cost reality
  • Capability / safety feedback loop - how AI-builder feedback feeds model + product + safety teams, cadence + precedents

Related roles

Sourced from

Ready to Generate Your Own Prep?

Drop your CV and a job description on the home page. A couple of minutes later you get a report with everything you need to land the job.

Generate my first prep