AI Data Engineer (Agent Platform)

Collective•San Francisco, CA

1d•Hybrid

About The Position

Collective is redefining the way businesses-of-one work by providing an integrated platform that handles business incorporation, accounting, bookkeeping, tax services, and community access. Our mission is to empower self-employed individuals to achieve financial independence and enjoy tax savings comparable to large corporations. We are backed by prominent investors and have been featured in major financial and tech publications.

Requirements

Product engineering experience with a focus on data.
Treat data models as part of the product, not a separate layer.
Strong opinions on schema design, naming, and consistency.
Comfortable identifying issues in production data and fixing them at the root.
Ownership mentality; ability to take vague, high-stakes problems and turn them into real systems.
Focus on outcomes over implementation.
Strong data experience, including data modeling, warehouses, and pipelines.
Ability to go from idea to architecture to production.
Experience building with LLMs, RAG, or agents in production.
Shipped systems that people actually rely on.
Effective communication with both technical and non-technical stakeholders.
Ability to drive alignment across teams with competing priorities.
Comfortable challenging existing systems and pushing for better approaches.

Responsibilities

Develop agent infrastructure, including retrieval, planning, and execution systems on top of real data, supporting multi-step reasoning and analysis workflows.
Integrate into internal tools to build an end-to-end analytics system with a natural language interface over the entire data stack, enabling users to run complex analyses without writing SQL.
Own and improve the data layer, collaborating with product engineering to design and evolve clean, consistent data models, and identifying/resolving inconsistencies at the source.
Build and maintain data pipelines, owning the reliability of the data infrastructure end-to-end, and debugging pipeline failures.
Make it easy to integrate new data sources into the data warehouse.
Ensure outputs are correct, consistent, and explainable by building evaluation loops, monitoring, and guardrails.
Eliminate ambiguity in metrics, definitions, and sources of truth, while handling permissions, sensitive data, and edge cases.
Drive adoption by working directly with teams across EPD, Member Operations, Legal, and more, driving alignment on data definitions and system usage, and making this the default way people interact with data.