Tiers predict what a company's interview loop scores for. The prediction is right most of the time. When it's wrong, you prepare for the wrong evaluation center — walk in with the evidence the tier rewards and discover the room is grading a different test.
This piece applies the deviation taxonomy from Issue #7 to eight roles on your active list, grouped by the deviation pattern they represent. Four patterns, two roles each. The roles are illustrations. The patterns are what you'll recognize in postings that aren't on this list.
Pattern 1 — AI-Native Company, Non-Agent-Behavior Mandate
The AI-native tier predicts evaluation centered on model-behavior judgment: delegation thresholds, supervision design, recovery from degraded states, prototype-led learning close to the model. That prediction breaks when the role's actual accountability is adoption economics or creative-product quality — when the AI is the medium, not the subject.
Read the posting's verbs. "Activation," "conversion," "retention," and "engagement" mean the company is hiring a growth leader whose product happens to run on AI. "Taste," "emotional resonance," "craft," and "creative direction" mean a creative-product leader whose material happens to be AI. AI fluency is table stakes either way, but it's not what the room is scoring.
OpenAI — Product Design Lead, Growth
Tier predicts: Model-behavior judgment, delegation architecture, trust calibration.
Actually screens for: Adoption-funnel ownership. The posting assigns discovery through activation, engagement, conversion, and retention for Codex and ChatGPT business users. Management plus hands-on product decisions. High confidence.
Lead with: Allē's shipped product-and-outcome chain (BCG Digital Ventures). It proves you own a full adoption funnel and ship against measurable results. Support with Agentic Labs and the Trust essay to hold AI credibility without letting them take over the conversation.
Checksum question: "For this role, which result is decisive — adoption and retention metrics, the quality of the human-AI interaction, or both? Who owns the tradeoff when they conflict?"
Landmine: Opening with delegation architecture and trust frameworks while leaving activation and retention unaddressed. The room hears "AI thinker" and wonders who ships the growth work.
Suno — Head of Product Design
Tier predicts: Agent-behavior design, human oversight patterns, AI-trust architecture.
Actually screens for: Creative leadership, taste, and consumer-product intuition. The posting weights people leadership, creative direction, product strategy, and execution across Product Design, Design Engineering, and UX Research equally. AI appears as context and medium. No delegation, supervision, permission, or recovery responsibility is named. High confidence.
Lead with: Equinox+ as consumer-product, multi-brand, rapid-build proof (BCG Digital Ventures). Support with Alibaba for design leadership at scale and Agentic Labs for current AI making. The Trust essay supports if the conversation moves toward creator control, but it shouldn't open.
Checksum question: "Is first-year success defined mainly by raising creative quality, scaling the design organization, or defining how users control AI-generated music?"
Landmine: Making governance and safe delegation the center of a conversation whose published criteria center on taste, joy, and consumer expression. The tool-fluency vs. product-fluency distinction from Issue #9 applies here. Suno wants someone who builds with AI, not someone whose primary claim is designing how users supervise it.
Pattern 2 — Growth-Stage Company Carrying Weight From Another Tier
The growth-stage tier predicts speed, direct customer learning, measurable outcomes, hands-on execution, coherence while scaling. That prediction fails when the role's product surface introduces consequence structures or infrastructure-governance problems that belong to the enterprise or regulated tier. The company still moves at growth speed. The evaluation adds a second layer the tier doesn't account for.
Two signals. First: the posting names a third party who experiences consequences of actions they didn't initiate — a patient affected by a provider's AI assistant, a developer whose production environment is modified by an agent. Second: the posting assigns responsibility for access, revocation, permissions, or safety alongside the usual shipping mandate. Either signal means the interview tests whether you can hold consequence and speed simultaneously.
Render — Staff Product Designer, Agent Experience
Tier predicts: Fast execution, customer proximity, system coherence, deployable outcomes.
Actually screens for: Human-agent interaction architecture across CLI, API, MCP, SDK, documentation, and web surfaces. The posting names handoffs, supervision, visibility, access, revocation, trust, safety, and governance. It also requires organizational alignment without formal authority — a growth-stage constraint layered onto an AI-governance mandate. High confidence.
Lead with: The Trust essay and Agentic Labs — Carrier IQ especially — for inspectable control patterns. Support with Alibaba for cross-surface coherence. TinyFish can ground traces, auditability, and reversibility in past-tense production context only.
Checksum question: "Can this role decide when an agent receives access, pauses, asks for help, or has access revoked — or does it make those decisions visible after another team sets them?"
Landmine: Offering only a fast infrastructure-product story. Without handoffs, permissions, supervision, and recovery across human and machine surfaces, you sound like a platform designer who hasn't considered what happens when the agent acts wrong. This role sits in the product-fluency category from Issue #9.
Headway — Staff Product Designer, Provider CRM
Tier predicts: Zero-to-one speed, ambiguity tolerance, direct customer learning, measurable adoption.
Actually screens for: Three-party consequence management. The role starts from a near-zero provider product surface and includes an AI front desk and scheduling system that acts on a provider's behalf while affecting patients. The posting emphasizes automation boundaries, operational complexity, and workflows across providers and patients. High confidence.
Lead with: Allē for paired provider/consumer product architecture and Thermo Fisher for exception-first regulated operations (both BCG Digital Ventures). Support with the Trust essay and Agentic Labs — Carrier IQ especially — for human review and approval patterns.
Checksum question: "Who decides what the AI front desk may do without confirmation when the provider delegates the work but the patient experiences the consequence?"
Landmine: Presenting zero-to-one speed as sufficient proof. The room wants to know you've thought about what happens when automation acts on behalf of one party and the other party bears the cost of a mistake. The three-party delegation problem Issue #8 flagged.
Pattern 3 — Enterprise Company Demanding Builder Speed or Organizational Invention
The enterprise tier predicts platform coherence, cross-product influence, durable standards, formal leadership structures, measurable impact at scale. That prediction fails in two directions. Sometimes the role demands hands-on technical making at a pace the enterprise tier doesn't anticipate. Sometimes the role's unit of accountability is the design organization's own operating capability rather than any product surface.
The first variant looks like a growth-stage IC role wearing an enterprise title. The second looks like a design-ops or transformation mandate wearing a product-design title. Both will misdirect your preparation if you default to the enterprise playbook.
Atlassian — Lead Product Designer, Loom
Tier predicts: Cross-product coherence, influence across mature systems, leadership that scales through established structures.
Actually screens for: Hands-on technical making, code contribution, AI prototyping, and multimodal human-agent experience shipping. Reports to Loom's Design Manager of Core Product. Assigns code contributions, pull-request fixes, AI-tool-accelerated prototyping, and mentoring. Cross-product work with Jira and Confluence is framed as hands-on design, not governance. High confidence.
Lead with: Agentic Labs as inspectable making evidence and the Trust essay for the human-agent interaction model. Support with Alibaba for enterprise-scale coherence. Don't let Labs carry a team-leadership claim — the role doesn't ask for one.
Checksum question: "Is 'Lead' here primarily senior hands-on product ownership, cross-product standards leadership, or leadership through other designers — and what's the expected split?"
Landmine: Leading with enterprise governance and executive influence while failing to show code-adjacent prototyping and personal shipping velocity. The posting's evaluation center is closer to what Issue #9's "Five Tests" called the technical-builder test than the function-builder test the enterprise tier would predict.
Capital One — Director, Product Design, AI in Experience Design
Tier predicts: Leadership of a customer-facing platform area, management layer, responsibility for coherence across products and teams.
Actually screens for: Organizational AI capability building. The Director works with the Chief Design Officer and owns roadmaps for AI tooling, AI-enabled design-process evolution, workforce transformation, and AI-centric upskilling across Capital One Design. The posting names Claude, Gemini, Claude Code, Stately, XState, and Windsurf as tools the Director should use hands-on. No customer-facing AI product or feature is owned by the role. High confidence.
Lead with: Alibaba for mandate creation, organizational influence, and cross-surface scale. Support with the Trust essay and Agentic Labs to establish that your view of AI-enabled design practice rests on product judgment and current making. TinyFish can supply past-tense technical currency or the bridge narrative only.
Checksum question: "What concrete organizational outcome does this Director own — tool adoption rates, changed design throughput, evaluation standards for AI-assisted work, or customer-facing AI quality? Which of those can the role decide versus recommend?"
Landmine: Pitching this as ownership of an AI banking product. The published mandate is transformation of how the Experience Design organization works. Issue #9's accountability classification would call this a function-builder test. Preparing for a product-area test means talking about the wrong surfaces entirely.
Pattern 4 — Scope Exceeding Its Apparent Container
A title can understate a role when the posting assigns product strategy, system ownership, cross-organizational coherence, or team formation beyond what the title normally carries. The deviation runs in one direction: the role is bigger than the label. The risk runs the other way: you position at the implied scope and the interviewer hears someone who's above the craft the posting demands.
The diagnostic is a split between two sections of the same posting. Requirements name specific craft standards, tools, surfaces, and implementation involvement — language that anchors the role at its posted level. Responsibilities assign strategy across a product suite, design-system ownership, team building, or cross-organizational patterns — language that belongs one or two levels above the posted title. When the responsibilities section would read naturally under a Director title but the requirements section would not, the role exceeds its container.
Stripe — Staff Product Designer, Risk
Tier predicts: A blend of enterprise platform coherence and regulated-tier consequence management.
Actually screens for: Cross-journey coherence across policies, operations, and shared Risk patterns while remaining part of Stripe's Product Design team. The posting asks the designer to translate systems, policies, and operational constraints into user-facing experiences, improve coherence across journeys and touchpoints, and contribute shared patterns for the broader Risk organization. Labeled "Team: Design" but no reporting line is named. High confidence on posting language, moderate on actual authority.
Lead with: Thermo Fisher's regulated, exception-first operational system (BCG Digital Ventures). It proves you hold both detailed craft and operational coherence across policies and exception flows — the dual demand this role creates. Support with Alibaba for cross-surface coherence and the Trust essay for evidence and approval mechanics.
Checksum question: "When policy requirements, legitimate-user friction, operational feasibility, and experience quality point in different directions, who closes the decision — and where does this Staff designer sit in that process?"
Landmine: Reducing the role to either design-system scale or generic regulated trust. The posting asks for both detailed craft and coherence across policies, operations, journeys, and shared Risk patterns. Choosing one frame and ignoring the other will sound like you read a different posting. Issue #8's reporting-structure probe matters here: whether day-to-day authority sits with a design leader or a Risk product leader changes your preparation.
CodeRabbit — Senior Product Designer
Tier predicts: Senior IC ownership of defined product areas, with function strategy and team formation held above the role.
Actually screens for: Suite-wide design strategy, design-system ownership across web, IDE, Git, and CLI surfaces, AI interaction patterns, model-adjacent prototyping, and team building — all under a Senior title. A separate posting confirms a Head of Design and a small product design team already exist. The scope reads closer to a Director mandate than a bounded Senior IC seat. High confidence on scope, moderate on actual authority.
Lead with: Agentic Labs — especially Carrier IQ — for AI interaction, trust, model proximity, and hands-on making. Pair with Alibaba because Labs alone don't prove organization building or cross-suite leadership. TinyFish for recent production context and technical currency only, past tense.
Checksum question: "How many product designers are on the team today, which responsibilities belong to the Head of Design, and does 'build a team' give this role approved hiring authority or mentoring and recruiting participation?"
Landmine: Opening as a Director-level manager above the craft. The posting's explicit gate is visual and interaction quality, rapid shipping, AI-native working practice, and close involvement in implementation. Leading with organizational architecture and skipping the making will sound like you're positioning for the Head of Design role that already exists.
Applying the Patterns to Roles Not on This List
Four steps.
-
Assign the tier. What does the company type predict about evaluation criteria?
-
Read the posting's verbs and accountability language. Does the role's unit of accountability match the tier's prediction? Growth verbs in an enterprise posting, creative-product verbs in an AI-native posting, organizational-transformation verbs in a product-design posting — each is a deviation signal.
-
Classify the deviation. Pattern 1: AI-native company, non-agent-behavior mandate. Pattern 2: Growth-stage company carrying consequence or governance weight from another tier. Pattern 3: Enterprise company demanding builder speed or organizational invention instead of platform governance. Pattern 4: Scope exceeding its title container. Most deviations you encounter will be variants of one of these four.
-
Reorder your evidence. Lead with what the role actually screens for. The tier's default evidence order is your fallback, not your plan.
If the answer contradicts the posting, that contradiction is the more valuable finding — it tells you what the hiring committee learned after they wrote the posting.
- Render's authority gap: The Agent Experience posting assigns domain accountability for trust, safety, and governance while requiring the designer to align the organization "without formal authority" — worth probing whether that constraint has changed since the posting went live.
- Capital One's missing product surface: No customer-facing AI feature appears in the Director posting, but Capital One's broader AI in Experience Design mandate may expand toward product ownership once organizational adoption reaches a threshold — ask about the roadmap.
- CodeRabbit's growing design team: At least two Senior Product Designers have publicly announced joining CodeRabbit since the posting appeared, which may shift the Senior role's team-building scope toward mentoring rather than founding.
- Stripe Risk reporting ambiguity: The posting labels the role "Team: Design" while embedding the designer in the Risk product team, and no manager is named — the first-conversation reporting-structure probe from Issue #8 should be the opening move.