Data Engineer
Description
Key responsibilities
• Build ingestion into the bronze layer for assigned sources: gateway and observability logs, productivity
tool admin APIs, AI-enabled SaaS usage, hyperscaler billing exports and reference data. Land raw and
untransformed, on a scheduled refresh, replayable if the downstream design changes.
• Work to the shared bronze landing contract so each tool is ingested once and serves both this program
and the parallel productivity initiative, rather than being integrated twice.
• Build the silver layer: typed, deduplicated and conformed to the canonical dimensions, refreshed
independently of any downstream publication schedule.
• Build gold marts carrying attribution method, attribution level, cost basis and provisional status alongside
cost and usage.
• Implement the attribution and allocation logic designed by the analysts, including precedence resolution
and ratio-based splitting of shared endpoint cost.
• Work within Unity Catalog governance — shared bronze and silver, separate gold marts with a recorded
owner per dataset — including permissions, lineage and cataloging.
• Implement data quality rules and monitoring: completeness, freshness and tag-coverage checks with
alerting, so pipeline problems surface before they reach a divisional invoice.
• Manage the volume impact of enabling caller-identity data in the cost and usage report, which multiplies
row counts by the number of calling identities per model.
• Work to the per-source cadence — daily where controls and anomaly detection depend on it, monthly
where they do not — within the team's existing CI/CD and promotion practices.
Essential skills and experience
• Advanced Databricks engineering: Delta Lake, medallion architecture, Databricks Workflows, Auto
Loader and incremental ingestion patterns.
• Unity Catalog to a governance standard — catalogs, schemas, permissions, lineage — not merely as a
place tables happen to live.
• Strong Python and PySpark, and strong SQL. Notebook-based development.
• Ingestion from REST APIs including pagination, throttling, incremental watermarks and credential
handling, plus cloud object storage across AWS, Azure and GCP.
• Performance and cost optimization of Spark workloads: partitioning, clustering, file sizing and cluster
configuration.
Tokenomics Program - Contract Role Descriptions | Page 7
• CI/CD for Databricks — asset bundles or equivalent — and Git-based development workflow.
• Able to work to an existing catalog structure and coding standard rather than introducing a parallel
approach.