Staff Engineer, Data Platform

Staff Engineer, Data Platform

About NationGraph

NationGraph is building the data and intelligence layer for the public sector.

More than 110,000 state and local government agencies across the U.S. independently publish information about:

How they operate

What they buy

Who they work with

What problems they are trying to solve

That information is fragmented across millions of websites, documents, databases, procurement systems, meeting records, and public records.

NationGraph turns that information into structured, connected, actionable intelligence for businesses selling to government.

Founded in 2024, NationGraph is dedicated to making uncommon knowledge common , because public data should actually be public.

The Role

We’re looking for a Staff Engineer, Data Platform to own one of the most important technical problems at NationGraph: turning the outside world’s fragmented government information into a proprietary data advantage.

This is not a traditional data engineering role focused on maintaining a warehouse or internal analytics.

You’ll own the technical ecosystem that:

Discovers external data

Acquires it reliably

Understands and extracts information from it

Normalizes and connects it

Validates its quality

Makes it available to NationGraph’s products and models

The scope starts with more than 110,000 independent state and local government agencies, but extends to federal data, Canada, and eventually public-sector information globally.

You’ll work across:

Data engineering

Distributed systems

Information retrieval

Data modeling

LLMs and agents

Applied ML

Entity resolution

Knowledge graphs

You’ll partner closely with Product, ML Research, and Infrastructure to determine both:

How we acquire data

What data NationGraph should have that nobody else does

What You’ll Do

Own our external data platform end-to-end

Design systems spanning discovery, acquisition, extraction, normalization, entity resolution, validation, storage, serving, and monitoring.

Establish the architecture and abstractions other engineers build on.

Map the world of government data

Develop a deep understanding of where government information lives.

Understand how it is published, how it changes, and how information across thousands of institutions can be connected.

Build systems for messy, real-world data

Work across government websites, APIs, procurement systems, PDFs, spreadsheets, meeting records, and public records.

Build for changing schemas, broken sources, conflicting records, and edge cases.

Use AI to rethink the traditional data stack

Work with our ML Research team to use LLMs, agents, and emerging models to:

Discover new sources

Understand unfamiliar schemas

Extract structured information

Resolve entities

Monitor data quality

Detect when sources change

Build proprietary data flywheels

Create systems where more data improves our models.

Use better models to discover and understand more data.

Continuously expand NationGraph’s underlying knowledge graph.

Set technical direction

Define the architecture for how NationGraph acquires and represents public-sector information.

Make decisions that will shape the platform over the next several years.

Help determine which technical investments create the strongest long-term data advantage.

You Might Be a Good Fit If

You’re an unusually strong engineer who genuinely enjoys working with data.

You’ve owned significant production data systems end-to-end.

You enjoy the detective work of making sense of unfamiliar, messy datasets.

You’re strong in Python, Go, or another systems/backend language.

You’re highly proficient with SQL.

You understand distributed data systems, including:

Orchestration

Idempotency

Backfills

Retries

Observability

Lineage

Failure recovery

You have experience with one or more of:

Large-scale external data

Crawling

Information retrieval

Entity resolution

Knowledge graphs

Document processing

You’re excited about using LLMs and modern ML as components of data infrastructure.

You care deeply about data quality, correctness, and reliability.

You have strong product judgment and can reason about what data is actually worth acquiring , not just how to acquire it.

You thrive in ambiguity and would rather create the architecture than be handed one.

We’re particularly interested in backgrounds spanning:

Alternative data

Quantitative research infrastructure

Search and crawling

AI data infrastructure

Knowledge graphs

Large-scale document processing

Data aggregation

None of these are requirements.

Our Engineering Stack

Backend: Python, Go, PostgreSQL

Infrastructure: Redis, Docker, Kubernetes

Frontend: React, TypeScript

AI / ML: LLMs, agents, proprietary models, and emerging frontier-model research

Our stack will evolve. At Staff level, you’ll help decide how.

Why NationGraph

Own a foundational problem

A large part of this architecture still needs to be invented.

You’ll have significant ownership over how NationGraph discovers, acquires, represents, and serves public-sector information.

Work on a genuinely hard data problem

There is no single API for American government.

There are tens of thousands of institutions, millions of sources, inconsistent schemas, and enormous amounts of information buried in systems never designed for machines.

Build a real data moat

We believe a major long-term advantage in applied AI will come from proprietary context and data.

Government contains enormous amounts of valuable information that is technically public but practically inaccessible.

Your job is to change that.

Work with exceptional people

You’ll work closely with the CEO, CTO, and a small engineering and research team.

The team has backgrounds spanning high-scale infrastructure, quantitative finance, AI, and startups.

Have real ownership

We move quickly.

We operate with very little bureaucracy.

Engineers have significant ownership over technical decisions and product outcomes.

If the idea of building the data infrastructure to map and understand how government works sounds exciting, we’d love to talk.