About the role
About the role
Nation Graph is solving a problem that has never been approached at scale – extracting quantitative intelligence from a fragmented, qualitative world. Across more than 110,000 public agencies, the most valuable signals are buried in PDFs, press releases, board meeting transcripts, town hall recordings, and government websites. Our mission is to do for the public sector what Bloomberg did for financial markets: transform disparate, unstructured information into structured, actionable intelligence.
As a machine learning engineer, you will be central to that mission. Where competitors rely on off-the-shelf LLMs, we build proprietary scrapers, data mining tools, and models from the ground up – because every output must be tied back to a specific, verifiable source. The technical challenges are substantial: disambiguating entity references across thousands of sources, parsing meaning from audio and large document corpora, and making principled decisions about data prioritisation across an enormous and largely uncharted landscape. For engineers who want to apply the rigour of quantitative finance to an entirely new domain, this represents a rare and genuinely open-ended problem.
Responsibilities
- Build and productionize end-to-end ML pipelines.
- Mine data from the web through large-scale crawling and scraping to power our models and insights.
- Transform unstructured text data into structured knowledge with NLP, entity recognition, and custom models.
- Build and improve text classification models to organize complex data.
- Optimize retrieval-augmented generation (RAG) systems used in our product.
- Drive our data strategy by identifying new data sources.
- Solve open-ended technical problems, teaching, learning, and iterating with the team.
- Work primarily in Python and SQL.
