Founding Engineer, Data & AI

About the role

Full-time. Founding team. New York City, in-person required. Reports to the Co-founder & CTO and works directly with both founders and domain experts.

About Pennant

Pennant is a YC-backed company building software for corporate governance, starting with proxy voting and company engagement.

Institutional investors, public companies and their advisors make consequential decisions using information scattered across filings, policies, research and conversations. We bring that information together so teams can understand the evidence, apply their own judgment and preserve why they made a decision.

Our ambition is a world model for corporate governance: a system that connects institutional knowledge, policies, decisions and outcomes. Getting there starts with reliable data and software customers trust in their daily work.

The role

Build and own the data systems that make Pennant useful and trustworthy.

This is a backend- and data-intensive applied AI role. You should be as comfortable debugging a production pipeline and evolving a database schema as evaluating a document-extraction model. You will build on an existing codebase, working with messy filings and customer documents and making the results dependable enough for customers to use. The work includes asynchronous jobs, versioned evidence and safe reprocessing, not just prompts and model experiments.

Your first mandate is one prioritized pipeline and its downstream use. You will establish what good looks like with domain experts, improve the system, and own it in production. As one of our first engineering hires, you will also help shape how we build, test and operate software.

What you'll own

- Build and improve ingestion and extraction pipelines for public filings, external feeds and customer documents. Choose structured sources, deterministic parsing or models based on the task. Make jobs safe to retry, prevent duplicate or stale-worker writes, recover from partial failures, and make missing or stale coverage visible.

- Design data models for the entities and relationships your pipeline needs. Preserve source evidence, effective dates and version history so past outputs remain reproducible as data, policies and models change. Distinguish verified facts, model outputs, human decisions and missing information.

- Build evaluations with domain reviewers using independently human-reviewed reference cases. Measure completeness, citation support, consequential errors and cases requiring human judgment, not just aggregate agreement. Treat model-generated references as proposals, not ground truth, and turn production failures into regression tests.

- Provide reliable services for research, policy application and recommendations. Make reprocessing safe: retain customer corrections, identify affected outputs and flag results that need renewed review.

- Protect customer information throughout ingestion, storage, evaluation and reprocessing. Respect tenant boundaries and permitted data uses. Do not assume one customer's documents or decisions can be reused for another.