We’re building an AI-ready data platform on top of our BigQuery warehouse and looking for a Senior Data Engineer to lead how AI interacts with data.
This is a data engineering role with real AI exposure — focused on structure, semantics, and control, not model training.
You will define how data is understood, accessed, and used by AI systems across the company.
We use AI-native workflows (Claude Code).
A core part of the system is structured instruction layers (skill.md) that define:
- how AI queries data
- how metrics are interpreted
- what is allowed vs restricted
You will design both:
- the data layer
- and the instruction layer that makes it usable by AI
Data Platform
- Build clean, reusable analytical data layers in BigQuery
- Move business logic from BI into the warehouse
- Define metrics, dimensions, and semantic consistency
AI Interaction
- Enable reliable AI ↔ data interaction
- Design schemas + instructions so AI produces correct outputs
- Test and refine real AI usage (not theory)
Access & Governance
- Implement data-layer access control (not BI-layer)
- Row/column-level security, role/attribute-based access
- Ensure consistent behavior across BI, AI, and internal tools
- Ensure metric reconciliation across different data sources
Prevent:
- sensitive data leakage
- shadow metric layer
- uncontrolled query cost
Automation:
- Replace manual data workflows with AI-driven processes
- Build agents for reporting, validation, and internal analytics
- Strong SQL + data modeling (BigQuery or similar)
- Experience building scalable data layers (not just pipelines)
- Understanding of metrics, semantics, and analytical correctness
- Hands-on exposure to LLMs / AI tools (practical, not theoretical)
- Ability to design systems where data is consumed by machines
- dbt / semantic layers
- BigQuery performance & cost optimization
- Finance / brokerage data
- Fine-grained access control implementations
- AI agent / tool-based workflows
- You will define how AI uses data in a real business
- Work on non-trivial problems (semantics, access, correctness)
- Build systems from first principles, not maintain legacy
- High ownership