AI Data Readiness Lead

Role overview

This position centers on making analytics trustworthy by owning the meaning of every metric the business runs on. As AI agents increasingly consume data and act on it, the role ensures metric definitions stay unambiguous, enforceable in code, and validated against ground truth so that wrong numbers are never applied confidently at scale. It is a governance-first role with real technical depth, where defining, implementing, and validating definitions takes precedence over pipeline building or new agentic reporting.

Responsibilities

- Establish canonical definitions for the metrics the business relies on, convene owners when competing versions exist, document disagreements, and drive to a single published decision with clear change rationale - Implement agreed definitions inside the semantic layer and data catalog so the system applies them automatically rather than relying on documentation, and retire superseded versions - Audit the existing reporting footprint, remove assets without an audience, and assign clear ownership to what remains - Build and maintain data quality checks covering freshness, uniqueness, referential integrity, and cross-system reconciliation, with failures routed to named owners or agents who act on them - Keep an inventory of AI agents that access company data and the definitions each depends on, then evaluate their output against known-correct answers and track accuracy, refusal, and error rates - Make governed data accessible and trustworthy for self-serve users, whether they query directly or through AI tooling

Requirements

- At least five years of experience in analytics, analytics engineering, or a closely related discipline - Strong SQL skills, including the ability to reverse-engineer undocumented transformation logic written by others - Direct ownership of a semantic or metrics layer in production using tools such as dbt, Cube, LookML, or equivalent, with accountability for what goes into it and why - Demonstrated success resolving conflicting metric definitions across functions and landing a final decision - Clear written communication, since most output is documentation others must trust without re-deriving it - Comfort deprecating and removing work that others have built - Experience evaluating LLM or AI agent output against ground truth - Experience developing or contributing to a data catalog and/or lineage tooling

Nice to have

- Familiarity with lakehouse architectures, Iceberg, Athena, Trino, or similar technologies - Exposure to audit readiness, SOX, or financial controls environments - Background in consumption- or usage-based business models where committed, consumed, invoiced, and recognized revenue are genuinely different numbers - Experience joining a function early, before established process existed