Postdoc in causal inference and language technology

We have the power of over 50,000 students and employees. Students who provide hope for the future. Employees who every day contribute to Linköpings universitet tackling the challenges of our time. Our core values rest on credibility, trust, and security. By being brave, thinking freely, and creating something new, we together, through large and small actions, create a better future. Welcome to apply for a job with us!

Your duties

You will be part of the research project Countering Bias in AI Methods in the Social Sciences, a collaboration between Institutet för analytisk sociologi (IAS) at Linköpings universitet and Institutionen för data- och informationsteknik (CSE) at Chalmers tekniska högskola, funded by WASP-HS. The position is located at IAS. The project lies at the intersection of causal inference and language technology (natural language processing, NLP).

A central challenge when drawing causal conclusions from observational data is adjusting for other contextual factors (confounders). Within the social sciences, these factors are often not measured directly but are instead documented in unstructured text. Using text to adjust for these factors is promising, but it carries a particular risk that we call treatment leakage: when the text also contains information about the treatment assignment, conditioning on the text can bias the causal estimation. Your task is to collaborate with our team at CSE, who lead the method development, and subsequently apply their methods to applied social science questions. The methods enable researchers to use text data for causal inference while ensuring this source of bias is detected and removed.

The work combines modern language technology with causal estimation. Current directions in the project include using large language models to remove treatment-predictive information from text, evaluating and comparing estimators that correct for bias, such as Design-based Supervised Learning and Prediction-Powered Inference, as well as applying techniques for mechanistic interpretability, for example sparse feature circuits and the SHIFT method, so that text classifiers disregard features associated with treatment assignment.

You will contribute to shaping the direction of the research, lead your own studies, publish in leading scientific fora, and collaborate closely with the project group, including serving as an assistant supervisor for the project's doctoral student and through collaboration with partners at University of Texas at Austin. Your primary focus will be on applying these methods to the impact of development projects in Africa. Secondarily, on similar causal problems.

As a postdoc, you will primarily conduct research. Teaching may also be included in the duties, however, at most one-fifth of the working time.

Your qualifications

To be eligible for employment as a postdoc, you must have completed a doctoral degree or have a foreign degree assessed to correspond to a doctoral degree. Your degree must be completed no later than the time the employment decision is made.

It is meritorious if your degree was completed no more than three years before the last application date for this position. If there are special reasons, a degree obtained earlier may be considered, for example, due to different types of statutory leave.

We are looking for you who have a doctoral degree in social science or computational social science, or in a related quantitative subject, as well as solid programming skills, for example in Python or R. You are also welcome to apply if your degree is in applied parts of statistics, computer science, machine learning, or language technology and has a focus on AI for Social Good or similar. Very good knowledge of English in speech and writing is a requirement, as the project is conducted in an international research environment. We also believe that you have solid education in at least one of the project's two pillars, causal inference or language technology, as well as a genuine interest in uniting the two. Experience in any of the following is highly meritorious: causal inference with observational data, text as data, large language models, mechanistic interpretability, or semiparametric and design-based estimation. Publications in leading scientific fora within social science, machine learning, language technology, statistics, or computational social science are highly meritorious.

You are a self-driven and collaborative researcher who enjoys working across disciplinary boundaries, takes independent responsibility for a research agenda, and communicates clearly with both computer scientists and social scientists.

Your workplace

You will work at Institutet för analytisk sociologi (IAS), which you can read more about here: https://liu.se/organisation/liu/iei/ias. At IAS, you will be part of professor Adel Daoud's research group, which works with artificial intelligence and causal inference for the social sciences, in close collaboration with the department for Data Science and AI at Chalmers.

About the employment

The employment is a fixed-term position for two years with the possibility of extension for a total employment period of at most three years. The position as postdoc is full-time.

Start date by agreement.

Security clearance may be conducted before a decision on employment is made.

Salary and benefits

Linköpings universitet applies individual and differentiated salary setting and is carried out in accordance with the applicable collective agreement (RALS/RALS-T) as well as the university's guidelines for salary formation.

Read more about benefits for employees https://liu.se/jobba-pa-liu/formaner

Union contact persons

Information about union contact persons, see https://liu.se/jobba-pa-liu/hjalp-for-sokande.

Application

You apply for this position by clicking the "Ansök" button below. Your application must be provided to Linköpings universitet no later than September 24, 2026. Applications received after the last application date will not be considered.

We welcome applicants with different backgrounds, experiences, and perspectives; it enriches and develops our operations. For us, it is fundamental to uphold everyone's equal value, rights, and opportunities. Read about our work at https://liu.se/artikel/lika-villkor/.

Welcome with your application!

Linköpings universitet has procured agreements and requests that we are not contacted directly by staffing and recruitment companies as well as sellers of job advertisements.