Data Engineer

We are looking for a Data Platform Engineer to join a 4-month contract opportunity based in Norman, Oklahoma. This role owns the storage, retention and pipeline-correctness workstream. This is data engineering, not cloud infrastructure.

What they will actually do

• Own the storage and pipeline backlog across the ingestion and processing repositories
• Event ordering and bitemporal modelling: separating event time from ingestion time, deterministic replay
• Atomic snapshot publication with schema admission and rollback
• Design the historical layer on object storage so it can be queried economically and answer as-of questions correctly - file format, partitioning, compaction, and whether the canonical layer moves onto a managed table format
• Reconcile the operational database and the historical store so a point-in-time question returns one answer rather than two
• Evidence immutability and retention: object-lock semantics, preventing overwrite of stored evidence, validated recovery paths
• Pipeline durability: idempotent runs, completeness receipts, restore testing
• Support production cutover in month 4 - this seat does not roll off early

Required

• 5+ years data engineering
• Python and SQL to a strong standard
• Practical experience with event ordering, idempotency and exactly-once or effectively-once publication
• Bitemporal or as-of data modelling - this is the single most important requirement
• Open table formats in production - Apache Iceberg, Delta Lake or Hudi. Not the name on a CV: partitioning strategy, compaction, snapshot isolation, schema evolution and time travel, and the judgement about when a table format earns its complexity
• Object-storage query engines - Athena, Trino, Spark SQL or equivalent - including why a query that was fast last quarter is slow now
• AWS data services: S3, RDS/Postgres, DynamoDB, Glue Data Catalog
• Uses AI coding tooling as part of their normal workflow

Useful but not required

• Managed Iceberg services (S3 Tables) or self-managed Iceberg on object storage
• dbt, Snowflake, Databricks or similar warehouse tooling - and a view on when a warehouse is and is not worth it
• Geospatial data (geometry, multipart and hole semantics)
• Content-addressed storage or evidence-preservation systems