Research Assistant / Doctoral Student (f/m/d) – AI & Historical Data at the SAFE Research Data[...]

Overview

The Leibniz Institute SAFE is seeking to fill the position of a Research Assistant / Doctoral Student (f/m/d) – AI & Historical Data at the SAFE Research Data Center.

The Leibniz Institute for Financial Research SAFE – Sustainable Architecture for Finance in Europe – promotes scientific research, transfer work to the public and independent policy advice on all aspects relating to the structure and functioning of the financial system. It aims to contribute to strengthening a sustainable and resilient financial system that drives innovation and serves the needs of the economy and citizens. Researchers from the fields of economics, law, and political science collaborate at SAFE. As part of a research project, we investigate the use of modern AI methods, in particular Vision Large Language Models, for the automated recognition and structuring of historical economic data.

The aim is to extend or potentially replace existing OCR-based approaches and, in particular, to significantly improve the recognition of complex layouts such as tables. As a use case, we work with the historical source “Bankenjahrbuch”, which, due to its structure, provides a challenging and highly relevant basis for evaluating modern methods.

The position is based at our institute’s Research Data Center and involves close collaboration with the Research Data Center of the University of Mannheim within the framework of and the Academy project “Financial and Corporate Research from a Long-Term Perspective”. You will work at the intersection of machine learning, data processing, and economic history, contributing to the development of new methodological approaches. The project offers the opportunity to work with unique historical data and to systematically evaluate modern AI methods in a relatively unexplored application area.

Position Details

Research Assistant / Doctoral Student (f/m/d) – AI & Historical Data

Type: 70% position, fixed‑term until December 31, 2028.

Start date: July 2026.

Remuneration: TV‑H E13 (collective agreement for the State of Hesse).

Responsibilities

• Evaluate current vision‑based models for text recognition and document structure understanding.
• Analyze and compare models with respect to accuracy and runtime performance.
• Create and prepare training datasets for model fine‑tuning.
• Develop and implement fine‑tuning and adaptation strategies.
• Compare modern approaches with existing OCR methods.
• Document results and contribute to scientific publications.
• Develop reproducible workflows and share results in line with Open Science principles.

Profile

• University degree (Master or equivalent) in computer science, data science, economics, or a related field.
• Interest in machine learning, especially NLP, computer vision, or multimodal models.
• Programming skills (especially Python) and experience working with data.
• Interest in experimental work, model evaluation, and methodological development.
• Structured and self‑driven working style.
• Interest in combining technical work with research‑oriented questions.
• Interest in economic history or digital methods in the social sciences is a plus.
• Initial experience with OCR, LLMs, or similar technologies is a plus but not required.
• Good English skills (written and spoken).
• Good German skills are an advantage.

Benefits

The position offers a high degree of flexibility and independence in shaping the research and technical approach, with a strong focus on experimentation and methodological development.

It provides the opportunity to start a doctoral project in finance, economics, or economic history.

Remuneration is based on the collective agreement for the public sector of the State of Hesse (TV‑H) at grade E13. An attractive occupational pension scheme through the German public sector pension fund (VBL) is also offered.

SAFE seeks to increase the share of women in research and thus strongly encourages female scholars to apply. We are also committed to promote the inclusion of persons with disabilities and invite them in particular to apply.