Staff Software Engineer - Full Stack

As a Staff Full-Stack Engineer on the Cloud Clients team, you'll own end-to-end delivery across the surfaces where dbt developers do their work: dbt Studio in the browser, the dbt VS Code extension, and the CLI workspace runtime that powers both. You'll help build dbt's AI-native developer experience, including agent workflows, diff review, streaming tool calls, charts and rich outputs for conversational analytics, and MCP apps that bring dbt projects directly into tools like Claude and Codex. This is a true full-stack role: you'll work across Python, TypeScript, React, and Kubernetes, with the expectation that you can take a feature from schema and service design through to the final pixel without a handoff

Responsibilities:

• Own parts of the CLI workspace runtime that backs both surfaces, including migrating users off our legacy IDE and enabling asynchronous, background agent runs.
• Design and build agent-facing experiences: streaming responses, transparent tool calls, diff review, and charting / rich output for conversational analytics.
• Extend dbt into other developer surfaces: MCP integrations that make dbt projects usable directly from AI clients like Claude and Codex.
• Lead well-scoped projects end to end: write the plan, partner with design and product, ship in two-week increments, and measure whether it worked.
• Share on-call and customer support for your surfaces, and mentor engineers earlier in their careers.

Requirements:

• 8+ years building production software, including senior-level ownership of projects that ran for months and involved several people.
• Deep Python experience plus exposure to TypeScript and React; you move across the stack without waiting for a handoff.
• Built and maintained something other developers depend on: an IDE, an editor extension, a CLI, an SDK, an LSP server, or a data tool.
• Comfort with the operational side of shipping APIs, Kubernetes, Postgres, observability and being on the hook when something breaks.
• Strong written communication: you can make a technical argument in a document and bring people along.
• Worked asynchronously as part of a fully-remote, distributed team.

Bonus points for:

• Shipped LLM- or agent-powered features to production, with real opinions on evaluation, latency, cost, and where models fall.
• Worked with the Language Server Protocol, Monaco, CodeMirror, or other code-editor internals.
• Hands-on dbt, SQL, or data warehouse experience (Snowflake, Databricks, BigQuery, Redshift).
• Migrated users off a legacy product surface and decommissioned it without breaking them.
• Go, Rust, or systems-level performance experience.