Duties and Responsibilities:
Data Architecture and Platform Ownership: Own the architecture for ingestion, transformation and analytics delivery. Decide the patterns for pipelines, storage layout and workspace structure, and hold them across products so we end up with one platform rather than several. Decisions are recorded with the reasoning and revisited on evidence
• A published architecture that every new pipeline is built against
• New product data onboarded onto the existing pattern rather than a bespoke build
• Platform decisions recorded with their reasoning and reviewed as things change
Standards, and How Agents Are Used on Data Work:
Write down the standards the team and the tooling both build from: schema conventions, transformation patterns, naming, and what qualifies as fit for reporting. Prove each standard with a working reference implementation. Set how coding agents are used on data work: what is specified first, what is always checked, and what is never taken on trust
• A data engineering playbook in the repository covering pipelines, naming and monitoring
• A reference implementation for every pattern the standard requires
• Generated transformation logic verified against known data before release, by a check that runs
Engineering Depth and Data Design: Pipelines and transformations are production code. They are designed, reviewed, tested and maintained, not assembled in a portal and left. Idempotency and safe replay, correct incremental logic, modelling for how the data will be queried rather than how it arrives, and clean separation between layers
• Design decisions defended at review on principle rather than preference
• Pipelines that replay and backfill correctly by design, not by luck
• Models other engineers extend without having to rebuild them
Running the Platform: Deployment, Monitoring and Cost: Everything reaches production through source control and a pipeline, with versioning, monitoring and a recovery path that has been tested rather than assumed. Own what the platform costs as volume grows, including the compute the agentic way of working consumes
• Pipelines deployed from source control rather than by hand, with drift detected
• Monitoring and alerting on every production pipeline, each with a named owner
• Cost per workload understood and acted on, and recovery proven by exercise
Data Quality and Governance: Put quality checks and validation at each stage so problems are caught where they enter rather than in a customer facing report. Keep metric definitions consistent across datasets, and work with security and infrastructure on access control, row level security and privacy
• Validation at every layer boundary, with failures visible and owned
• One agreed definition per business metric, used by every dataset that reports it
• Access control and row level security agreed and evidenced rather than assumed
Analytics Enablement: Architect the semantic layer- models, data marts and shared datasets. Guide analysts and report developers on modelling, performance and reuse, so business users can trust and interpret the numbers consistently
• Shared datasets reused across reports rather than duplicated per report
• Report and refresh performance held to agreed targets as volume grows
• Analysts able to build without needing an engineer for every change
Leading the Data Function: Provide technical leadership to the data engineers: guide solution design, review work, resolve the hard problems, and coach the team in lineage, quality and observability. You work in a team that spans more than one location and lead it as a peer rather than through a chain. You will be judged substantially on what the engineers around you can do because you were here
• Named engineers visibly more capable, with review load moving off you over time
• A plan for scaling the workload and onboarding further engineers
• Peer assessment from engineers and architects outside your reporting line