Data Engineer

Stealth Startup; Amsterdam; On-site; EUR 70,000 - 110,000; Full time

Job summary

Stealth Startup in the Netherlands is seeking a Data Engineer to own the path from raw OSINT data to the representations our AI agents reason over. You will ingest and transform data at billion-row scale and build provenance-aware structures.

You’ll work with Python, PySpark, Iceberg, ClickHouse and OpenSearch, collaborating with AI engineers and end users. Experience with large-scale, text-heavy and imperfect datasets is valued; comfort with ambiguity and an AI-native mindset is essential.

Qualifications

• Experience building data pipelines at scale (billions of rows).
• Strong Python and SQL skills.
• Knowledge of data modeling and provenance.
• Experience with OSINT or text-heavy datasets is a plus.

Responsibilities

• Ingest and transform raw data into analyzable representations.
• Build graph/temporal/provenance-aware data structures.
• Ensure data quality and traceability from source to derived facts.
• Collaborate with AI engineers and end users to ship infrastructure.

Skills

Python

PySpark

Big data

Tools

Iceberg

ClickHouse

OpenSearch

Job description

We’re a fast-growing AI scale-up working with analysing social data to predict human behaviour at scale.

We’re building agentic AI systems that need to reason over the real world and the real world produces terrible data.

Billions of records. Unstructured text. Conflicting sources. Missing information. Constantly evolving datasets. And in our world, an AI agent confidently reasoning over the wrong representation of that data can be worse than having no answer at all.

Following our Series A, we’re looking for a Data Engineer to join our team in the Netherlands.

This isn’t traditional data engineering. You willl own the path from raw open-source intelligence data to the representations our AI agents and analysts actually reason over.

That means:

• Ingesting and transforming data at billion-row scale
• Building agent-ready graphical, temporal and provenance-aware data structures
• Making messy, incomplete and adversarial data usable
• Ensuring every derived fact can be traced back to its source
• Working across Python, PySpark, Iceberg, ClickHouse and OpenSearch

You’ll work directly with AI engineers, behavioural scientists and end users. Sometimes you’ll build a POC quickly to solve an immediate operational problem. Other times you’ll turn what we learn into infrastructure that becomes part of our core platform.

We’re particularly interested in engineers who have worked with large-scale, text-heavy and imperfect datasets and know what happens when data systems start to genuinely hurt at scale.

Experience with OSINT, graph databases, temporal data, sovereign infrastructure or secured/ regulated environments is a plus, but not essential.

What matters is engineering depth, comfort with ambiguity and an AI-native way of working.

If building the data substrate that autonomous AI systems ultimately trust sounds like your kind of problem.