Senior Analytics Engineer - Decision Intelligence

Overview

Deepgram is building its decision layer, one governed set of numbers that people and AI agents can act on without re-deriving them. This senior technical IC role owns the modelling, semantic layer and agent-facing surface on a new team, as one of the first or second engineers in.

What you'll do

• Model the gold layer, with conformed entities for customer, account, project, product and contract, including resolving identity across Salesforce, the console, Stripe and the billing system.
• Own the semantic layer by authoring models that define metrics once so a dashboard, a notebook and an agent return the same number, versioning them and retiring superseded ones.
• Build the agent-facing surface, including tools, permissions and MCP surfaces that let analyst agents navigate data without inventing incorrect joins.
• Make correctness checkable through assertions on grain, uniqueness and cross-system reconciliation routed to a named owner, and evaluate agent output against known-correct answers, tracking accuracy, refusal and confidently-wrong rates separately.
• Answer questions from sales, product and finance, and leave behind a model that makes the next version of that question cheaper.

What you'll need

• AI-native practice, demonstrated through daily use of AI as a way of working that has changed what you attempt.
• Production-grade SQL and Python, with code that ships on a schedule and has been found broken by your own instrumentation.
• Semantic or metrics layer ownership of dbt, Cube, LookML or equivalent, not just use, including having changed a definition people depended on.
• Modelling judgement that reasons about grain and keys before writing a query, and distinguishes a data defect from an undecided business definition.
• Comfort working on work-in-progress infrastructure with no ticket queue or inherited process.
• Clear writing for definitions and contracts others must trust without re-deriving them.

Nice to have

• Usage-based or consumption business models, where committed, consumed, invoiced and recognised revenue are different numbers.
• Real-time billing and consumption systems.
• Lakehouse architectures such as Iceberg, Athena, Trino or similar.
• Contributions to open-source data or AI projects.

Details

• Location: USA, remote.
• Expectation to actively use and experiment with advanced AI tools in everyday work.