Data science is the broadest field of the AD7 ICT competition in content. Annex II groups its duties into eight categories that run from data engineering and statistical modelling to semantic technologies, high-performance computing, MLOps, analytics and data governance.
This guide covers field 4 of EPSO/AD/429/26, the AD7 competition for IT professionals (notice in the Official Journal C/2026/2425, 6 May 2026). Applications closed on 10 June 2026 and the testing period is 2 December 2026. For the dates, the four fields side by side and the format of every test, see the AD7 ICT overview.
Data science in numbers
| Item | Value |
|---|---|
| Places (successful candidates sought) | 184 |
| Applications | 1,917 |
| Applications per place | ≈ 10 |
| Candidates whose EUFTE is scored (in principle, 1.5 × places) | up to ≈ 276 |
| Testing period | 2 December 2026, online, remotely proctored |
| Field-related MCQ | 30 questions, 40 minutes, second language, pass mark 15/30 |
With 184 places and 1,917 applications, data science drew about ten applications per place, the competition average. Applications count submissions, not people who will sit the tests, so the real ratio on 2 December is lower. The field MCQ ranks everyone, and in a field this wide the test can reach topics outside your day-to-day specialism.
What the job involves, according to the notice
Annex II lists eight duty categories: The notice says successful candidates may be given any of these duties, alone or combined.
- Data engineering. Technical design and management of data ecosystems and distributed computing architectures; ETL/ELT processes, orchestration frameworks and scalable pipelines for large, heterogeneous datasets.
- Statistical modelling and quantitative analysis. Statistical models, econometric techniques, time-series forecasting; feature engineering, outlier detection, sample stratification and Monte Carlo simulation for inference and model validation.
- Data architecture and semantic technologies. Enterprise data architectures using knowledge graphs, ontologies and linked open data; metadata standards, multilingual thesauri and taxonomy generation with graph databases and RDF.
- High-performance computing and optimisation. HPC for computation-intensive processing, distributed clusters, GPU-accelerated analytics, parallel algorithms and optimisation.
- Data integration and APIs. Integration solutions, microservices and APIs, data virtualisation, real-time synchronisation and automated data quality monitoring with anomaly detection.
- MLOps and automation. Automated machine learning pipelines, model versioning, automated training, hyperparameter optimisation, A/B testing, production deployment, containerised ML, CI/CD and automated monitoring and retraining.
- Analytics and visualisation. Advanced analytical frameworks, real-time dashboards, self-service analytics and common business analytics tools (Power BI, Qlik).
- Data governance and communication. Data stewardship, ownership, reference data management, lineage, metadata management and data quality; aligning standards with stakeholders; guidance, training and expert advice.
Who was eligible
Field 4 is the only one with its own list: data science or data analytics; data, information or knowledge management; artificial intelligence or computational linguistics; mathematics; statistics; and any other discipline with a study focus on data science. That last route needs evidence: a recognised data-science track named on your diploma or transcript, coursework worth at least 30 ECTS (one semester of full-time study), or a thesis on data science. With a degree outside the list you needed eight years of relevant experience, one fewer than in the other three fields.
The general scale is the same across fields: a three-year degree plus seven years of relevant experience, a four-year degree plus six, or a five-year degree plus five. An advanced degree (master's, PhD or equivalent) in a listed area satisfies the education requirement of the four- and five-year routes. Applications are closed, so this section matters only if you want to check what the competition asked.
What the field test is likely to cover
EPSO publishes no syllabus for the field MCQ. The list below is our reading of the Annex II duties: the best map available, not an official list. The test is specific to the field you chose, and it is the only one that ranks you.
- Pipelines and architecture. ETL versus ELT, orchestration, distributed processing, data lakes and warehouses at scale, and where semantic layers (ontologies, RDF, knowledge graphs) fit.
- Statistics and modelling. Time series, econometrics, outliers, stratified sampling, Monte Carlo methods and how to validate a model.
- Machine learning in production. Model versioning, retraining, hyperparameter optimisation, A/B tests, deployment strategies and monitoring for drift.
- Integration and APIs. APIs, microservices, data virtualisation, real-time synchronisation and data quality checks.
- Analytics and visualisation. What dashboards and self-service platforms are for and how real-time analytics differs from batch reporting.
- Data governance. Stewardship, ownership, lineage, metadata and data quality, the vocabulary of managing data as an asset.
How to use the time before 2 December
Put most of the remaining time into the field MCQ, because that is the test that ranks you. Work through the duty categories above one at a time, and for each ask whether you could answer a scenario question about it in your second language, since the test is taken in that language. Thirty questions in forty minutes leaves about eighty seconds each, so practise reading technical English, French or German at speed, whichever your second language is.
Then clear the gates. The three reasoning tests are pass/fail: verbal needs 10 out of 20, and numerical and abstract need 10 out of 20 combined, so a few timed sessions in your first language are worth more than perfecting them. Last, read the EUFTE source document in the days before the test. It is published in 24 languages and is built on Regulation (EU, Euratom) 2023/2841; the essay assesses written communication, not field knowledge.
Sharpen your reasoning tests
Verbal, numerical and abstract reasoning are common to every EPSO competition. Practise them under realistic timed conditions, free.
Start practising → Free sample cards · every answer explainedAfter the tests
Results are sent only at the end of the competition and EPSO has not announced a date. The selection board fills the reserve list for the field in ranking order, up to 184 names, shown alphabetically. Being on the list is not a job offer: the institutions recruit from it through their own interviews. The grade guide explains what AD7 means.
Frequently asked questions
What is the data science field of the AD7 ICT competition?
Field 4 of EPSO/AD/429/26, an open competition for AD7 administrators in the EU institutions. It seeks 184 successful candidates in data science, out of 782 across four fields, and a candidate may apply for only one field.
How many people applied for data science?
1,917 applications for 184 places, about 10 per place, according to EPSO's competition page for the field. That counts applications submitted, not people who will sit the tests.
Which test decides the ranking?
The field-related MCQ test: 30 questions in 40 minutes in your second language, pass mark 15 out of 30. The three reasoning tests are pass/fail gates, and the EUFTE essay is pass/fail and scored only for the top-ranked candidates, in principle up to 1.5 times the places, here about 276.
What evidence counts for a 'study focus' on data science?
The notice accepts any one of three: a formally recognised specialisation or dedicated track in data science identified on the diploma or another official document; data-science coursework equal to at least 30 ECTS, or one semester of full-time study; or an academic thesis focused on data science.
When are the tests?
EPSO's competition pages give 2 December 2026 as the testing period. The tests are taken online under remote proctoring, and the exact slot and instructions arrive in your EPSO candidate account.
Can I still apply?
No. Applications and identity documents were due on 10 June 2026 at 12.00 Brussels time, and supporting documents on 1 October 2026. Check the AD7 overview and EPSO's page for what comes next.
Keep reading
- The AD7 ICT overview — all four fields, the dates, the odds and the test format in one place.
- AD7 ICT infrastructure — field 1: 204 places, 1,082 applications.
- AD7 ICT project management — field 2: 228 places, 4,421 applications.
- AD7 Clouds and networks — field 3: 166 places, 548 applications.
- The field-related test explained — multiple-choice, written test and essay.
- The EUFTE essay — how it is marked, with a worked example.
Whichever field you chose, the reasoning tests are common ground. Start with a few timed questions and see where you stand.