Staff Software Engineer

dbt Labs dbt Labs · Data AI · India · Remote · Engineering

Staff Software Engineer to architect and build the durable memory substrate for agentic analytics workflows, storing and exposing organizational decision memory (meaning, intent, rationale, history) for humans, agents, and applications. This greenfield role focuses on building the core Context Platform, designing Decision Memory schemas, owning context storage systems (graph, vector, time-based), building read/write/query APIs, and ensuring permission-aware, auditable, interoperable, and portable context access.

What you'd actually do

  1. Prototype apt technical solutions and find best fits for the context engine. Architect and build the core Context Platform
  2. Design schemas and primitives for Decision Memory and enterprise context
  3. Own context storage systems (graph, vector, event/time-based)
  4. Build read/write/query APIs used by agents, products, and external apps
  5. Design permission-aware, auditable context access

Skills

Required

  • Significant experience building distributed systems, data platforms, or infrastructure
  • Comfort operating in ambiguous, greenfield problem spaces
  • Deep expertise in data modeling and schema design
  • Experience designing shared platforms used by many teams
  • Strong instincts around APIs, contracts, and backward compatibility
  • Ability to reason about systems, not just components

Nice to have

  • Experience with knowledge graphs, metadata systems, or search/retrieval systems
  • Experience building systems with governance, auditability, or compliance requirements
  • Familiarity with dbt or modern analytics stacks or developer tooling

What the JD emphasized

  • durable memory substrate
  • agentic analytics workflows
  • Decision Memory
  • context storage systems
  • permission-aware, auditable context access
  • interoperable, portable, and zero-lock-in by design

Other signals

  • building the durable memory substrate that powers agentic analytics workflows
  • stores not just metadata, but _meaning_: decisions, intent, rationale, and history
  • makes it safely accessible to humans, agents, and applications