Director, Applied AI & Strategy

What matters most

• Find high-value workflow interventions

  The director analyzes complex workflows across business functions and Technology & Operations, prototypes agentic improvements, refines them with business feedback, and estimates implementation costs. Prepare to explain how you would identify friction, demonstrate a useful prototype, and build a case for production.

• Deliver trustworthy production agents

  The posting calls for production-grade agents with guardrails, evaluations, and observability for accuracy, reliability, and safety. It also requires hands-on LLM work, including multi-agent orchestration and accurate retrieval of curated unstructured content. Prepare to distinguish a working demonstration from a system ready to operate.

• Integrate across enterprise systems

  Agents must connect to enterprise resources through MCP and to applications, including legacy systems, through APIs. The role also requires architecture across complex SaaS ecosystems and composable toolsets. Prepare an integration approach that addresses dependencies without locking the solution to one tool.

• Make productivity gains measurable and governed

  The director partners with Legal & Business Affairs and Cyber Security on AI governance, while tracking hours reclaimed, cost savings, efficiency gains, and ROI. Prepare to connect safeguards and measurement to a specific workflow intervention.

• Turn AI developments into an executable roadmap

  The role translates leadership goals into a technical roadmap, prioritizes high-ROI work, manages stakeholders and scope, audits emerging tools, and gives leadership actionable media-relevant “Buy, Build, Ignore” assessments. Prepare to explain how evidence, cost, and business fit would shape those decisions.

Likely challenges to think through

• Speed versus production readiness

  A likely challenge is preserving rapid prototyping while adding the evaluations, guardrails, observability, and cost estimates needed for production. Prepare a staged approach that shows what a prototype must prove before scope and investment increase.

• Integration ambition versus system constraints

  Connecting agents to complex SaaS ecosystems, enterprise resources, and legacy applications could make a promising workflow difficult to implement reliably. Prepare to discuss how you would map interfaces and dependencies, test retrieval quality, and keep components replaceable.

• Efficiency versus governance and credible ROI

  An inferred trade-off is that faster workflows may not produce defensible savings if measurement is weak or governance concerns limit deployment. Prepare to describe baseline metrics, post-deployment checks, and how you would involve Legal & Business Affairs and Cyber Security without treating approval as an afterthought.

Topics to prepare

• Agent architecture and orchestration

  Review when advanced prompting, a single agent, or multi-agent orchestration is appropriate; sketch decision paths and failure modes for a hypothetical internal workflow. The posting requires production-ready LLM agents and multi-agent orchestration.

• Unstructured-content retrieval

  Diagram how content would be curated, enriched, retrieved accurately, and supplied as LLM context; explain how you would assess retrieval quality at scale. The posting explicitly requires RAG frameworks and hands-on RAG at scale and/or LLM fine-tuning.

• Enterprise integration and composability

  Sketch boundaries between agent logic, complex SaaS workflows, MCP-connected resources, and API-connected legacy applications; identify where components could be replaced. These integrations and composable toolsets are explicit responsibilities.

• Production assurance and governance

  Define plausible accuracy, reliability, and safety failure cases and map each to a guardrail, evaluation, or observable signal. The role builds production-grade agents and partners with Legal & Business Affairs and Cyber Security on governance.

• Productivity economics and delivery strategy

  Work through a hypothetical prioritization using implementation costs, hours reclaimed, savings, and efficiency gains; outline a proof-of-concept-to-production roadmap. The posting calls for ROI tracking, budgets, scope management, and translating leadership goals into execution plans.

• Media-relevant AI decisions

  Practice a concise Buy, Build, Ignore assessment of a hypothetical AI tool for a media-workflow friction point, stating assumptions and audit criteria. The role calls for media-industry insights and actionable leadership assessments; the particular workflow is only a practice scenario.

Skills to revise

• Analyze and prototype a workflow

  Map a hypothetical cross-functional workflow, select one agentic intervention, and build a small functional automation using a posting-named AI-augmented development tool such as Cursor, Replit, or Claude. Rehearse incorporating business feedback and estimating production effort; rapid prototyping is explicitly required.

• Implement retrieval-backed agent behavior

  Using sample unstructured documents, build a small RAG agent that curates source content, retrieves relevant passages, and checks whether responses reflect the retrieved context. This practices the posting's hands-on agent and RAG requirements without implying a specific employer dataset.

• Connect agents to enterprise interfaces

  Create a small illustrative agent integration through MCP and an API adapter; document input validation, error handling, and how a legacy application could be isolated behind the adapter. The posting explicitly names MCP, APIs, and legacy-system integration.

• Test and observe agent outputs

  Assemble a small test set with expected outcomes and unsafe or inaccurate cases; run evaluations, record failures, and describe the signals you would monitor after deployment. This exercises the stated guardrails, evaluations, observability, and safety responsibilities.

• Present an execution decision

  Prepare a one-page hypothetical business case with cost assumptions, an hours-reclaimed calculation, stakeholder checkpoints, a scoped roadmap, and a Buy, Build, Ignore recommendation. This practices the role's budgeting, ROI, stakeholder-management, and leadership-advisory duties.

• Map evidence to the core requirements

  Prepare a one-page evidence map with specific examples, where available, of enterprise integration, complex SaaS architecture, workflow automation, production LLM agents, RAG at scale or fine-tuning, and stakeholder leadership. The posting specifies these areas; do not claim experience you cannot substantiate.

• Rehearse a production project walkthrough

  Choose a relevant project you can discuss. Prepare a short account of the workflow, your hands-on and leadership decisions, architecture, integrations, stakeholder feedback, production outcome, and measured results. Separate your contribution from the team’s.

• Practice diagnosing a productivity workflow

  As a preparation exercise, select a hypothetical media-business workflow and sketch its current steps, bottlenecks, and an agentic intervention. State assumptions rather than presenting the scenario as an A+E project; identify what business feedback would change your design.

• Draw an agent and retrieval architecture

  Prepare a diagram showing where prompting, multi-agent orchestration, and curated unstructured-content retrieval fit. Include proposed MCP access to enterprise resources, API connections to applications or legacy systems, and choices that keep the toolset composable.

• Plan a rapid prototype demonstration

  Outline a small functional automation you could explain or demonstrate, including how AI-augmented coding tools helped, what the prototype proves, what feedback it needs, and what remains before production. The posting emphasizes rapid delivery and refinement, not a specified interview exercise.

• Define a production-readiness review

  Prepare a checklist of guardrails, evaluation cases, observability signals, and failure handling for accuracy, reliability, and safety. Show where Legal & Business Affairs and Cyber Security would review governance decisions; do not assume their specific policies.

• Build a roadmap and ROI sketch

  Draft a phased path from proof of concept to production with scope decisions, stakeholder checkpoints, cost-estimate assumptions, and a budget range. Explain how you would track hours reclaimed, cost savings, efficiency gains, and ROI to prioritize workstreams.

• Prepare an industry assessment and employer questions

  Create a brief hypothetical Buy, Build, Ignore assessment for an AI tool relevant to media workflows, with technical and functional audit criteria. Bring questions about which workflows are priorities, available integration constraints, governance review, and how success is measured.