Open positions

We took part in the , with application deadline of September 21, 2026. The application deadline has now passed and you can no longer apply for these positions, but the information for the positions is still available below e.g. for candidates preparing for the interviews. We were looking for two postdoctoral researchers for 2-3 year positions and a doctoral student for the full duration of PhD studies (3-4 years), starting as soon as possible.

1. Postdoctoral researcher in Prior elicitation and specification

   We develop methods for eliciting tacit human knowledge and expertise, so that we can embed this knowledge into statistical models. We are looking for a postdoctoral researcher to join the Flexible priors for flexible models (2024-28) project to contribute to this general research theme, with flexible exact topic. We expect previous expertise on probabilistic modelling with demonstrated publication record in key ML conferences (AISTATS, NeurIPS, ICML etc) and/or statistics journals. Possible directions are listed below, but you can also explain broader researcher interests:

   • Representation and elicitation of arbitrary beliefs using flexible machine learning models, similar to our recent works in and .
   • Elicitation, encoding and use of flexible forms of prior knowledge. Instead of priors on model parameters, we are interested in functional priors, priors expressed over predictive distributions (, ), and use of priors within active learning setups like Bayesian optimization ()
   • Work that addresses the key challenges in use of prior elicitation in practice (): High-profile applications where use of human expertise is critical, concrete elicitation workflows and software, etc.

   • Quantification of LLM knowledge and uncertainty using elicitation techniques, taking the technical solution of further

      

2. Postdoctoral researcher and PhD student in AI-assisted workflows in healthcare

   The project joins forces of and the to create a nation-wide foundation model for healthcare decision-making. The project will both create a dedicated foundation model that is trained on electronic health records to provide individual-level information on population health and develop AI-assisted workflows for optimal use of that model and all complementary information sources in addressing concrete questions in healthcare policy, medical research and clinical practice. These workflows will typically include general-purpose LLMs, the foundation model that grounds the decisions in empirical data, and human experts (e.g. medical doctors or policymakers) as central component.

   We are looking for a postdoctoral researcher and a doctoral student to join this large initiative, focusing primarily on the question of how to best use the foundation model across diverse use-cases and what kind of additional components are needed for addressing concrete information needs. For a postdoctoral researcher we expect previous research expertise that aligns with some of the key directions below. Doctoral students are expected to express clear interest in specific themes, while demonstrating methodological competence in machine learning and/or agentic AI. Possible concrete topics include:

   • AI-assisted scientific research and agentic workflows, to contribute to the general architecture design together with dedicated research software engineers
   • Active learning and Bayesian experimental design, e.g. for recommending diagnostic tests based on the current evidence or for identification of control patients for medical research tasks
   • Decision-making and user models, to quantify both human and LLM knowledge and to combine it with the information provided by the foundation model 
   • Interpretable AI and human-AI collaborative interfaces, to create concrete pilot systems that integrate and demonstrate the work across the whole project

We will always consider exceptional students interested in doctoral studies and strong candidates for post-doctoral researcher positions. Please contact the group leader directly if interested. Some general topics of interest the group is working on right are listed below, and you are advised to relate your interests to these topics. Check also for possible open positions in artificial intelligence research in several research groups in the Helsinki area.

1. Probabilistic methods, with specific interest on efficient approximate inference (MCMC, Laplace, variational approximation) and flexible models (GPs, flows, diffusion models etc). Examples of our recent activities and work in this area: 

• Research Council of Finland projects Efficient Riemannian Inference (2022-24) and Computationally Efficient Inference on Riemannian Embedding Manifolds (2023-25)
• Yu et al. , AISTATS, 2024
• Luu et al. , NeurIPS, 2024
• Williams et al. , PGM, 2024
• Kusmierczyk et al. , NeurIPS, 2019

2. Collaborative AI and human modelling, with specific interest in eliciting and using tacit human knowledge. We develop theory and methods for eliciting human beliefs and knowledge as probabilistic quantities, to be used e.g. as prior information in scientific statistical models, or for explaining human behaviour. Examples of our recent activities and work:

• Research Council of Finland project Flexible priors for flexible models (2024-2028)
• Mikkola et al. , NeurIPS, 2024
• Mikkola et al. , Bayesian Analysis, 2024
• de Souza da Silva et al. , JMLR, 2023
• Hartmann et al. , UAI, 2020

3. Machine learning for science, with specific interest in virtual laboratory as domain-agnostic platform for assisting scientific research. We work e.g. on Bayesian optimization, physics-informed machine learning, neural operators, and on the general question of how to best use machine learning solutions within scientific discovery. The work can be done in context of several application areas, ranging from ultrasound physics to food science. For instance, we work on AI-assisted design of novel food production processes via use of plant and fermentation-based proteins, as part of the . Examples of our recent activities and work:

• Research Council of Finland projects Food Innovation and Diversification to Advance the Bioeconomy (2025-27) and Sustainable Industrial Ultrasonic Cleaning (2023-25), Business Finland project Virtual Laboratories for pharmaceutical R&D (2023-25), and Transcranial AI-powered ultrasound computed tomography (2026-29) funded by Jane and Aatos Erkko Foundation.
• Klami et al. , Data-Centric Engineering, 2024
• Gharib et al. , Mechanical Systems and Signal Processing, 2025
• Mikkola et al. , AISTATS, 2023