Project Positions for Postdoctoral Researchers and Research Fellows in ICT (Helsinki, Finland)

Aalto University and the University of Helsinki are looking for Postdoctoral Researchers and Research Fellows in multiple areas of ICT, including: artificial intelligence and machine learning data science privacy and security computational health human-computer interaction. We are committed to fostering an inclusive environment with people from diverse backgrounds, and researchers from underrepresented groups are particularly encouraged to apply.

The deadline for applications is January 30th, 2021 at 11:59 PM (23:59 UTC+02:00). By applying to this call, organized by Helsinki Institute for Information Technology HIIT and Finnish Center for Artificial Intelligence FCAI, you use one application to apply to positions for both of our hosting institutions, Aalto University and the University of Helsinki. Aalto University and the University of Helsinki are the two leading universities in Finland in computer science and information technology. Both are located in the Helsinki Metropolitan area, and the employing university will be determined by the supervising professor. Please see below for additional information on the application process. (In the section entitled “How to Apply”)

The Helsinki Institute for Information Technology (HIIT) and the Finnish Center for Artificial Intelligence (FCAI) bring together the world-class expertise of Aalto University and the University of Helsinki in joint research centers and programs. There are multiple research groups from either Aalto University or the University of Helsinki participating in this open call. Recruited researchers will work in one of the participating groups, or in some cases, a shared position may also be possible.

Research Group Projects

These projects are not directly funded by HIIT, but by individual researchers or research groups:

H1: Machine Learning in Precision Oncology

Cross-disciplinary research group of Prof. Sampsa Hautaniemi focuses on discovering causes for drug resistance in cancer patients and effective treatment options to overcome them. We have strong track-record in computational method development for data from patient samples (e.g., and translating research results to clinical practice (e.g., We coordinate one of the largest ovarian cancer precision oncology projects in the world funded by EU Horizon 2020 (DECIDER; In DECIDER our focus is to discover effective treatment options for ovarian cancer patients who do not respond to chemotherapy anymore. More information and all our publications are available at our group website:

We are looking for post-doctoral researchers with strong experience in data science, machine learning or mathematical modeling, and interest in applying computational methods to cancer research. Basic knowledge of cancer biology or genetics is considered as a plus. Examples of the possible projects are: computationally effective methods for very large datasets; mathematical modeling of tumor evolution; integration multi-omics cancer data; image analysis of histopathological images; and causal analysis of single-cell data.The Hautaniemi group belongs to the Research Program in Systems Oncology (ONCOSYS; ONCOSYS consists of internationally acclaimed basic, translational and clinical researchers who share the vision of using cutting-edge measurement technology, real-world data and AI in cancer research in order to discover effective diagnostic, prognostic and therapeutic approaches. If you enjoy doing science in an interdisciplinary and collaborative environment, consider applying to us.

Supervisor: Sampsa Hautaniemi (University of Helsinki)
H2: Machine Learning for Health (ML4H)
Recent years have witnessed accumulation of massive amounts of health data, enabling researchers to address a range questions such as: how to allocate healthcare resources fairly and efficiently, how to provide personalized guidance and treatment to users based on real-time data from wearable devices, or how to use genomic data to understand disease or antibiotic resistance. Central challenges in ML4H include integrating noisy data from heterogeneous data sources, going beyond correlation to learn about causality, interpreting the models, and assessing the uncertainty of predictions, to name a few. We tackle these by developing models and algorithms which leverage modern machine learning principles: Bayesian neural networks, deep latent variable models, interactive machine learning, attention, reinforcement learning, and natural language processing. Our ongoing interdisciplinary projects include: analysis of nationwide healthcare register data, mobile health, genomics, antibiotic resistance, and epidemiology. Successful applicants are expected to have outstanding skills in machine learning, statistics, applied mathematics, or a related field. The focus of the position may be tailored based on the applicant’s interests to either methodological or application-driven research questions, and examples of our both kinds of recent research can be found in

Supervisor: Prof. Pekka Marttinen (Aalto University)
H3: Statistical Genetics and Machine Learning
The emergence of genomic sequencing and other molecular profiling methods together with the digitalization of health care data collections have created unprecedented opportunities for better treatment options. INTERVENE is an international and interdisciplinary consortium that seeks to leverage these vast, but underused data resources to generate clinically actionable knowledge for improved understanding of diseases and treatment options tailored to individuals, by analyzing altogether 1.7 million sequenced genomes with longitudinal clinical data. We are looking for an outstanding  postdoctoral researcher to join the interdisciplinary and international group of researchers in INTERVENE. From an applicant we expect a PhD in statistical genetics or a related field, or a PhD in a methodological field (e.g., machine learning, applied mathematics, statistics) coupled with a strong interest to apply the skills in the genetics application. More information about the project:

Supervisor: Prof. Pekka Marttinen (Aalto University)
H4:Bayesian Workflows for Iterative Model Building and Networks of Models
Statistical analysis is critical when it comes to obtaining insights from data. Despite the practical success of iterative Bayesian statistical model building, it has been criticized to violate pure Bayesian theory and that we may end up with a different model had the data come out differently. In this project, we formalize and develop theory and diagnostics for iterative Bayesian model building. The practical workflow recommendations and diagnostics guide the modeller through the appropriate steps to ensure safe iterative model building, or indicate when the modeller is likely to be in the danger zone.

Sometimes the user wants also to build and compare models of different complexity or based on different assumptions.  Similar workflow ideas can be used to support also analysis of networks of models, making it easier to illustrate similarities and important differences. By making applied scientific research and data analysis more reliable and reproducible, our understanding of the world and decision-making will be improved. Related paper and video

Supervisor: Aki Vehtari (Aalto University)
H5: Bayesian Inference and Machine Learning
We develop Bayesian models and computationally efficient inference algorithms for challenging modelling problems, and are looking for candidates with demonstrated contributions in machine learning and/or Bayesian statistics to join the team. We are currently working on, for instance, (a) computationally efficient Riemannian manifold MCMC with interesting extensions for scalable inference for Bayesian deep learning and variational approximations (with Mark Girolami, University of Cambridge), (b) Gaussian processes and deep learning models for ultrasound propagation in complex structures and real-time control of acoustic fields for levitation (with Ari Salmi, University of Helsinki), (c) prior elicitation (see, and (d) data-efficient learning via domain adaptation and physically motivated model constraints. We are looking for candidates interested in any of these topics, and you can also propose other topics aligning with the research directions of the group.

We expect you to have strong publication record in machine learning, statistics or signal processing, and interest in continuing fundamental research in this research area. We offer you the opportunity to work on the core algorithms together with other machine learning researchers and a chance to demonstrate the methods in context of interesting application domains with unique data provided by our collaborators, while enjoying in a friendly and pleasant working environment.

See for more information.

Supervisor: Arto Klami (Helsinki)
H6: Deep Generative Modeling for Precision Medicine and Future Clinical Trials
We are looking for a postdoc to develop novel probabilistic machine learning methods for large-scale health datasets from biobanks, clinical trials and/or single-cell sequencing experiments. This project develops novel deep generative modeling methods to (i) predict adverse drug effects using longitudinal/time-series data from large-scale biobanks and clinical trials, and to (ii) harmonize large-scale health data sets for AI-assisted decision making to revolutionize future clinical trials. Methodologically this project includes e.g. VAEs, GANs, Bayesian NNs, domain adaptation, Gaussian processes and causal analysis. The work will be done in collaboration with research groups from the Finnish Center for Artificial Intelligence, and the novel methods will be tested using unique real-world data sets from our collaborators in university hospitals and big pharma company.

See our recent work:





For more information, see

Supervisor: Assoc. Prof. Harri Lähdesmäki (Aalto)
H7: Bayesian Deep Learning for Continuous-Time Dynamical Systems
Recent machine learning breakthroughs include black-box modeling methods for differential equations, such as Gaussian process ODEs [1] and neural ODEs. These methods are particularly useful in learning arbitrary continuous-time dynamics from data, either directly in the data space [1] or in a latent space in case of very high-dimensional data [3]. We are looking for a postdoc to join our current efforts to (i) develop efficient yet calibrated Bayesian methods for learning such black-box ODE models, (ii) develop neural ODEs to learn arbitrary dynamics of high-dimensional systems (e.g. in robotics, biology, physics or video applications) using a low-dimensional latent space representation, and (iii) further developing these methods for reinforcement learning and causal analysis.

See our recent work:







For more information, see

Supervisor: Assoc. Prof. Harri Lähdesmäki (Aalto)
H8: Deep Representation Learning – Foundations and New Directions
Applications are invited for a PhD position and a Postdoctoral position in deep representation learning, broadly construed. Topics of particular interest include:

(1) Generative Models
(2) Graph Neural Networks
(3) Neural ODEs/PDEs/SDEs, Deep Equilibrium Models, Implicit Models
(4) Differential Geometry/Information Geometry/Algebraic Methods for Deep Learning
(5) Learning under limited data, distributional shift, and/or uncertainty
(6) Bayesian Methods, Probabilistic Graphical Models, & Approximate Inference
(7) Fair, diverse, and interpretable representations
(8) Off-policy reinforcement learning, inverse reinforcement learning, and causal reinforcement learning
(9) Multiagent systems and AI-assisted human-guided models
(10) Learning on the edge (i.e., learning under resource constraints)
(11) Applications in physics, computer vision, drug discovery, material design, synthetic biology, quantum chemistry, etc.
(12) Quantum Machine Learning for structured spaces

Representative publications:

(1) John Ingraham, Vikas Garg, Regina Barzilay, and Tommi Jaakkola. Generative Models for Protein Design. NeurIPS (2019).
(2) Vikas Garg, Stefanie Jegelka, and Tommi Jaakkola. Generalization and Representational Limits of Graph Neural Networks. ICML (2020).
(3) Vikas Garg and Tommi Jaakkola. Solving graph compression via Optimal Transport. NeurIPS (2019).
(4) Vikas Garg, Lin Xiao, and Ofer Dekel. Learning small predictors. NeurIPS (2018).
(5) Vikas Garg, Cynthia Rudin, and Tommi Jaakkola. CRAFT: Cluster-specific assorted feature selection. AISTATS (2016).
(6) Vikas Garg, Adam Kalai, Katrina Ligett, and Steven Wu. Probably approximately correct domain generalization. AISTATS (2021).
(7) Vikas Garg and Tommi Jaakkola. Predicting Deliberative Outcomes. ICML (2020).

Supervisor:Prof. Vikas Garg (Aalto)

H9: Machine Learning for Human-AI Collaboration
The Cognitive computing research group is seeking 1-2 postdoctoral researchers to conduct research in Machine learning for Human-AI collaboration. The candidate will work on machine learning methods that allow new types of man-machine interactions in which machines can learn directly from human behavior and physiology. The methods will be developed and applied for physiological sensing and brain-computer interfacing data. The position provides an opportunity to contribute both to machine learning for physiological computing and ground-breaking human-computer interaction research. The candidate will join world—class research environment and will be part of the Cognitive computing  research group ( The starting date is negotiable, but the positions are filled when suitable candidates are found.


The work involves research-related activities, including conducting theoretical and applied research, data analysis, writing research articles, participating in and presenting research at academic conferences, and teaching-related activities. The ideal candidate has a strong background in the intersection of machine learning and human-computer interaction, a PhD degree in computer science or related field, and publications in top-tier venues. Sole understanding of modern machine learning methods and good programming skills are required.

Supervisor: Tuukka Ruotsalo (Helsinki)
H10: Foundations of Computing
We are hiring postdoctoral researchers working on the foundations of computing. We welcome applicants working in all areas of theoretical computer science, broadly interpreted, including e.g. algorithmics and algorithm engineering, computability and computational complexity, computational logic, optimization, cryptography, computational geometry, natural computation, and foundations of distributed, parallel, and quantum computing.

We offer the possibility to participate and take initiative in leading-edge research in a young and growing research environment with 10 professors and their teams working on foundational topics in the Helsinki area at Aalto University and the University of Helsinki (*). The postdoctoral researcher positions are full-time research positions for a duration of one year, with the possibility of extension to a second year by mutual consent. Travel funding is available for travel permitted by the pandemic situation. Participation in teaching of advanced courses and thesis instruction is possible and encouraged, with 5-10% allocation of the total working time.



Chris Brzuska (Aalto)

Parinya Chalermsook (Aalto)

Petteri Kaski (Aalto)

Mikko Koivisto (Helsinki)

Juha Kontinen (Helsinki)

Sándor Kisfaludi-Bak

Pekka Orponen (Aalto)

Alexandru Paler (Aalto)

Jukka Suomela (Aalto)

Jara Uitto (Aalto)
H11: Human Sciences – Computing Interaction (HSCI)
The HSCI research group seeks to develop sound data-centric research practices in the humanities and social sciences. If you’ve ever wanted to tackle complex methodological issues in a domain with both complex research questions as well as datasets, but crucially also clear, meaningful and impactful end goals, then this position is for you. We are currently searching for 3-5(!) methodology-focused postdocs for 2-4(!) years to strengthen our interdisciplinary research group. The candidates are expected to have a background in a subfield of computer/data science or statistics suitable for this, such as in applied probabilistic modelling, explainable machine learning, deep learning, large-scale machine learning, natural language processing, or HPC or other workflow/tool design and development.

If selected, depending on your own interests, possible environments in which you might test your skills range from analysing massive digitized historical datasets to understanding what kinds of rhetorical group strategies competing thought communities utilize in online debates. For more on the broader collaboration environment, see

Supervisor: Professor Eetu Mäkelä (University of Helsinki)
H12: AI Technologies for Interaction Prediction in Biomedicine
Several positions are open in a project with the aim of developing new AI techniques for predicting interactions of objects. Such problems are prevalent in society, found in numerous applications, such as various link prediction problems in social networks, recommendation systems, and healthcare.
The project will tackle several key challenges, including improving explainability, efficient use of resources, and reliability of the predictions. The technologies will be rigorously evaluated in practical applications in biomedicine such as predicting drug combination responses in cancer (Julkunen et al., Nature Communications, 2020). The positions are affiliated to a large multi-year research grant funded by Academy of Finland.

Julkunen, H., Cichonska, A., Gautam, P., Szedmak, S., Douat, J., Pahikkala, T., Aittokallio, T. and Rousu, J., 2020. Leveraging multi-way interactions for systematic prediction of pre-clinical drug combination effects. Nature communications, 11(1), pp.1-11.

Supervisor: Juho Rousu (Aalto University), Tero Aittokallio (FIMM), Tapio Pahikkala (University of Turku), Antti Airola (University of Turku)
H13: FoTran – Found in Translation
FoTran is an ERC-funded project that focuses on representation learning in multilingual neural language and translation models. We develop and study large-scale sequence-to-sequence models trained on hundreds of languages to see how complex neural networks are able to pick up semantic knowledge from raw data. We look for highly motivated researchers who want to push the limits of multilingual NLP and who would be interested in system development, interpretability of neural networks and downstream applications in language technology. The ideal candidate would have a background in computational linguistics or computer science with experience in NLP and deep learning. We mainly work with pyTorch and common frameworks for neural machine translation and sequence-to-sequence models such as OpenNMT-py, fairseq and others. Our project makes use of high-performance computing and heavily depends on GPU-based computation. Experience with SLURM and HPC infrastructures is an advantage for the position. We have access to extensive IT infrastructures provided by CSC Finland and are selected as pilot project for the new European supercomputer LUMI. Our team is international and combines people with various backgrounds demonstrating the interdisciplinarity of our work. The position would initially be appointed for 18 months with a possibility of an extension depending on upcoming funding. More information about our research unit is available at and

Supervisor: Jörg Tiedemann (Helsinki)
H14: Lifecycle Support of Machine Learning Applications
The project focuses on MLOps, the DevOps type of practices applied to machine learning systems. The essence is to develop methods and techniques for all lifecycle activities, not just for the initial development of a machine learning model. The work is part of the IML4E ITEA-funded project (, which started recently.

We are looking for a PostDoc with an interest in scientifically approaching the challenges of developing, deploying, and operating machine learning systems. An ideal candidate would have good competence in both machine learning and software engineering. The detailed work will be refined with other research partners and with the background of the candidate.

In our team, we are running multiple projects in the area of AI practice.  Collaboration between the different projects and between the industrial participants is important.

Supervisor: Prof Jukka K Nurminen (Helsinki)
H15: Aalto Human-Computer Interaction and Security Engineering Lab
There are four (4) project positions open within the Aalto Human-Computer Interaction and Security Engineering Lab:

Project: Enjoyable security

We are looking for post-docs interested in transforming science in security and HCI. Background and interest in measuring user experience, modeling, human-computer interaction, computer science or social and behavioral sciences is required.

Project: Science of human-computer interaction

Are you frustrated in the lack of science in the field of human-computer interaction (HCI)? We are looking for post-docs interested in transforming science in HCI. Background and interest in measuring user experience, modeling, human-computer interaction, computer science or social and behavioral sciences is required.

Project: Multitasking and productivity tools

We are looking for post-docs interested in understanding productivity tools and multitasking. Background and interest in measuring user experience, modeling, human-computer interaction, computer science or social and behavioral sciences is required.

Project: Mixed methods HCI and security research

We are looking for post-docs interested in pushing the envelope in mixed methods HCI and security research. Background and interest in measuring either qualitative methods or quantitative methods, and interested to learning new methods, user experience, modeling, human-computer interaction, computer science or social and behavioral sciences is required.

Examples of work done in the Aalto Human-Computer Interaction and Security Engineering Lab can be found at the website for the group Please contact me at the email address about these positions during January 2022.

Supervisor: Janne Lindqvist (Aalto)
H16: Eco-Evolutionary Control Theory
Evolution connects all living organisms and is the common thread across biology. Organisms evolve to better survive in their environments and to adapt to new challenges. This leads to complex dynamical scenarios, which are presently understood only in a limited way. Understanding evolution is one of the most intriguing scientific topics due to its ability to unify often seemingly disjoint fields of biology. Furthermore, quantitative understanding of evolution is a prerequisite to successfully combat pathogens, pests and loss of biodiversity. Predictability and control of evolving populations is an emerging topic of high scientific interest1,2 and vast translational potential in applications such as vaccine and therapy design. This project will develop eco-evolutionary control theory. For example, we will find out what are the key determinants of controllability of a multi-species bacterial community, and apply the results to therapy optimisation. These projects are at the interface of microbial evolution, statistical physics, and information theory. A highly quantitative background and a burning interest in evolution are required.

Keywords: stochastic optimal control, eco-evolutionary dynamics, evolutionary theory, drug resistance.

  1. Lässig M, Mustonen V, Walczak AM (2017) Predicting evolution. Nat Ecol Evol 1(3):1–9.
  2. Lässig M, Mustonen V, (2020) Eco-evolutionary control of pathogens. PNAS, 117 (33), 19694-19704

The ideal candidate will have experience in reinforcement learning, stochastic optimal control theory and evolutionary theory. However, as our research is cross-disciplinary it is possible to contribute coming from several different fields. Thus, we welcome applications from exceptional candidates, with a highly quantitative background, from other fields.

For more information, please contact Prof. Ville Mustonen

Supervision: Professor Ville Mustonen (University of Helsinki); This is a collaborative project with the Lässig Group at the University of Cologne and the Särkkä Group at Aalto University.
H17: Computational Biology
All proliferating cells are exposed to natural selection. In nutritionally challenging environments, the selection acts on cellular phenotypes that are determined by the metabolic network of the cells. The metabolic network is encoded in the genome loci of metabolic enzyme encoding genes. In AdaGe project we seek to find out how this organization of the metabolic enzymes in the genome and the metabolic network structure interplay in cellular adaptation. We are looking for a Postdoctoral Researcher to work on modelling of metabolism and adaptive evolution.

The ideal candidate will have a PhD in computational biology, biotechnology, bioinformatics or similar field and has experience in (constraint-based) metabolic modelling and (linear) optimization. Additional background in eukaryotic genetics, NGS data analysis, or microbiology or similar would be helpful.

For more information, please contact Prof. Ville Mustonen

Supervision: Professor Ville Mustonen (University of Helsinki) and Dr. Paula Jouhten (VTT).
H18: AI algorithms for quantitative biology
Biology is being transformed by the arrival of Big Data. As it stands, further biological insights are limited by the lack new bespoke AI algorithms that can be applied on these data. These algorithms must be theory aware of the respective scientific problem so as to be interpretable. Only this way will AI powered analyses lead to sustained scientific progress in biology and other natural sciences. Here we develop theory aware, interpretable, AI for automated growth law determination form massively parallel phenotyping data. The project builds a foundation towards AI guided microbial growth control.

The ideal candidate will have experience in machine learning methods e.g. reinforcement learning and symbolic regression. However, as our research is cross-disciplinary it is possible to contribute coming from several different fields. Thus, we welcome applications from exceptional candidates, with a highly quantitative background, from other fields.

For more information, please contact Prof. Ville Mustonen

Supervisor: Professor Ville Mustonen (University of Helsinki)
H19:Variable Selection with Missing Data with Applications to Genetics
Our current ability to measure massive numbers of variables, for example, in biological sciences (millions of DNA variants), in personal health applications (thousands of time points in activity monitoring) or in customer behavior modeling (credit card usage history) call for efficient ways to separate the important variables from unimportant ones.

We are looking for a Postdoc to work on realistic variable selection problems where variables have been measured incompletely and where we have access only to statistical summaries of the underlying data. Our previous work include the Bayesian approach implemented in the FINEMAP algorithm ( In this project, our motivating example is the world’s largest study on migraine genetics [] where these new variable selection methods are needed to assess which genetic variants are biologically connected to the migraine susceptibility. The project includes both theoretical (statistical) and empirical (simulations and genetic data analysis) components.

Research group pages:

Supervisor: Matti Pirinen (Helsinki) [link:]
H20: Reconstructing Crisis Narratives for Trustworthy Communication and Cooperative Agency
Join a collaborative team of PhDs, Postdocs and academic researchers working on a project jointly conducted between Aalto University and the Finnish Institute for Health and Welfare (THL). The project seeks to analyze and reconstruct crisis narratives using mixed-methods, combining qualitative research for narrative inquiry with computational data analytics of crisis discourses in news and social media to understand global pandemics. Candidates will work at the intersection of Human-Computer Interaction (HCI), design research, computational social sciences, and public health for critical societal impact. We expect the candidates to have backgrounds in computer science and/or the social sciences.

Potential duties and tasks may include the following (to be conducted as part of the research team):
1. Examining the narratives emerging in crisis-related communication using qualitative research methods across various data sources including organization communications, news/media coverage, and social media exchange among diverse publics.
2. Automating content analysis for narrative work using suitable machine learning techniques for Natural Language Processing (NLP) such as Conversation Analysis, Content Classification, and/or Sentiment Analysis. This includes collecting and curating datasets, devising suitable methodologies, setting up the research infrastructure and tools, and a pipeline for data extraction, analysis and validation.
3. Representing and visualizing crisis narratives to support understanding and collaborative sensemaking among key stakeholders and diverse publics. These tools should support browsing, searching, and exploring information for selected crisis themes and narratives. Work in this area includes not only developing prototypes of visualizations, but also conducting design research, user experience (UX) evaluation, and pilot assessment of such tools.

The candidate is not expected to master all these domains, but work closely with a multi-disciplinary research team to lead design and development efforts, while learning and contributing to ongoing work in specific research areas of interest.

Research Areas: Computational Social Science, Machine Learning, Human-Computer Interaction (HCI)

Supervisor: Nitin Sawhney, Professor of Practice, Department of Computer Science, Aalto University
H21: Sustainable ICT
ICT has had tremendous effect on our world, bringing new kids of services and enhancing old ones. In terms of sustainability, ICT helps, for example, to reduce green house gas emissions in other sectors, digital solutions support environmental protection and adaptation to climate change. Yet, at the same time the use of ICT is growing at a huge pace. The traffic on the Internet and mobile networks grows year after year, and new and larger data centres are built all over the world. At the same time the performance of ICT hardware has increased extremely fast. As an example, a smart phone today has the same computing performance as the super computers in the 90’s. ICT services consume increasing amounts of energy and natural resources are needed for building new devices.

The goals of this project are to understand the reasons why the use of ICT is growing and what can be done to lower the direct impact of the ICT sector on our environment. Growth of ICT can be attributed to new services but also to how services are built, and how little optimisations are done to make these digital services more resource aware.

Supervisor: Professor Jukka Manner, Aalto

H22: Probabilistic modelling and Bayesian machine learning

I am looking for a postdoc or research fellow to join the Aalto Probabilistic Machine Learning Group, to work on new probabilistic models and inference techniques. I have exciting research topics available around the following areas, and I am also open to new suggestions: (1) simulator-based inference, for combining first-principles models with learning from data, (2) Bayesian deep learning, (3) Bayesian reinforcement learning and inverse reinforcement learning, (4) with multi-agent modelling, and (5) privacy-preserving machine learning and synthetic data generation. Can be theoretical or applied work or both; the group has excellent opportunities for collaboration with top-notch partners in multiple applications, including user interaction, health, genomics, and neuroscience. Links:

Supervisor: Samuel Kaski, Aalto

FCAI positions for Winter 2022 Call

Finnish Center for Artificial Intelligence FCAI is a research community that develops new types of AI that can work with humans in complex environments, and help renew industry and society. FCAI is built on a long track record of pioneering machine learning research. Currently over 60 professors contribute to our research.

AI-Assisted Virtual Laboratories

We at FCAI believe AI will transform the way research and development work is carried out in a broad range of disciplines. This requires a coordinated effort in identifying the central tasks and challenges of the research process itself, and development of collaborative AI methods for assisting in them. FCAI is driving this in the form of the Virtual Laboratory concept, developing the basis of AI-assisted support for research carried out in laboratories combining automated physical measurements and computational simulations. We are building the foundations and the AI methods needed for this, focusing in particular on techniques that can be applied across multiple fields of science. In parallel, we are establishing the first pilot Virtual Laboratories, to both steer the research and provide highly visible demonstrators. To contribute to this transition, you can apply for different kinds of roles, and work either on the underlying principles and methods (Topics F0, F6-F8; or for specific techniques and problems choose F9-F16) or directly on a specific Virtual Laboratory (Topics F1-F5).

F0: Virtual Laboratories Assisted by Collaborative AI: From Foundations to Practice
Many of the key elements in the AI-assisted research process, as carried out in a Virtual Laboratory, are common across disciplines. For instance, a common recurring task is the choice and design of the experiment to be carried out next, and for efficient AI-assistance we need AI that can understand the goals and motivations of the researcher.

We are looking for people interested in these fundamental questions related to Virtual Laboratories and AI-assisted research. You will identify commonalities to build the foundational principles, work on better computational methods for addressing recurring challenges while working in collaboration with others who focus on a specific Virtual Laboratory, and will work towards a concrete software environment that supports several example cases. The long-term goal is to make it easy for everyone to run domain-specific Virtual Laboratories.

We are looking for people with different kinds of profiles, with demonstrated excellence and strong desire to contribute to (a) the foundational basis of collaborative AI-assistance, (b) the required probabilistic machine learning techniques, and/or (c) open-source scientific software.

Supervision: Professors Arto Klami (University of Helsinki), Samuel Kaski (Aalto University), other virtual laboratories

Keywords: Virtual laboratory, AI-assisted design
F1: Virtual Laboratories: Multi-level Simulation for Sustainable Autonomy
To study future sustainable mobility, FCAI is building Sustainable Mobility and Autonomous Systems Virtual Laboratory. The virtual laboratory will allow studying effects of autonomous traffic starting from control of individual vehicles, to their environmental effects such as pollution and noise, as well as their socio-economic effects. The virtual laboratory will integrate several simulators including an autonomous vehicle simulator, as well as other simulators modeling relevant phenomena.

A central challenge in the integration is the exchange of information between the individual simulation models with different parameterizations. We approach this as an AI challenge where parameters of all simulators are inferred jointly from pools of data for each model. This will require efficient data-driven inference procedures.

More information about the topic >>

Supervision: Profs. Laura Ruotsalainen (University of Helsinki), Ville Kyrki (Aalto University), potentially with other supervisors

Keywords: Multi-level simulation, sustainability, autonomous vehicles, simulator-based inference
F2: Virtual Laboratories: Closing Simulation – Real World Gap
To study future sustainable mobility, FCAI is building Sustainable Mobility and Autonomous Systems Virtual Laboratory. The virtual laboratory will allow studying effects of autonomous traffic starting from control of individual vehicles, to their environmental effects such as pollution and noise, as well as their socio-economic effects. The virtual laboratory will integrate several simulators including an autonomous vehicle simulator, as well as other simulators modeling relevant phenomena.

When relying on simulation models for data-driven analytics, a central issue is the reality gap, the difference between a simulation model and the real world. In practice, simulation parameters need to be inferred often from scarce real-world data. However, in addition to this calibration problem, the simulation model is unlikely to capture all real-world phenomena. Thus, addressing the sim-to-real problem requires also determining this residual gap in order to compensate for it. This problem requires data-efficient probabilistic methods that are simultaneously expressive.

More information about the topic >>

Supervision: Profs. Ville Kyrki (Aalto University), Laura Ruotsalainen (University of Helsinki), potentially with other supervisors

Keywords: Sim-to-real problem, reality gap, data efficiency, autonomous mobility
F3: Virtual Atmospheric Laboratory
We are looking for postgraduate students and postdoctoral researchers to build Virtual Atmospheric Laboratory (VILMA). The objective of VILMA is to model atmospheric molecular level processes efficiently and to understand the underlying mechanisms and causal connections. VILMA will combine first-principles quantum chemical and other simulations and probabilistic machine learning/artificial intelligence (ML/AI) models with interactive visualization.

Examples of fundamental ML/AI topics are: probabilistic emulator / predictive regression models for atmospheric processes (Lange 2021; Lumiaro 2021), randomization methods for interactive visual data exploration (Puolamäki 2020), advanced statistical methods for ML/AI (Savvides 2019), explainable AI (Björklund 2019), and Bayesian optimization (Todorovic 2019).

You will work in multidisciplinary team of computer and atmospheric scientists. You should have basic knowledge of ML and related mathematics. We will consider applicants with backgrounds in computer science, atmospheric science, physics, and chemistry. Knowledge of natural sciences is considered an advantage, but specific prior knowledge of atmospheric processes is not required.


More information:

Supervision: Profs. Kai Puolamäki (University of Helsinki), Patrick Rinke (Aalto University), Hanna Vehkamäki (University of Helsinki), Theo Kurtén (University of Helsinki)

Keywords: Machine learning, probabilistic modeling, simulator-based inference, natural sciences
F4: Virtual Laboratories: Synthetic Psychologist
Theories in psychology are increasingly expressed as computational cognitive models that simulate human behavior. Such behavioral models are also becoming the basis for novel applications in areas such as human computer interaction, human-centric AI, computational psychiatry, and user modeling. As models account for more aspects of human behavior they increase in complexity. The Synthetic Psychologist Virtual Laboratory broadly aims to develop and apply methods that assist a researcher in dealing with complex and intractable cognitive models. For instance, by developing optimal experiment design methods to help with model selection and parameter inference, or by using likelihood-free methods with cognitive models. This virtual lab will also encourage avenues of research relevant to cognitive modeling and AI-assistance which can be pursued in collaboration with other FCAI teams and virtual laboratories. We are looking for excellent candidates who are excited by cognitive models, Bayesian methods, probabilistic machine learning, and in open-source software environments, in no order of preference.

Supervision: Profs. Luigi Acerbi (University of Helsinki), Andrew Howes (Birmingham),  Samuel Kaski (Aalto University), Antti Oulasvirta (Aalto University)

Keywords: Virtual laboratory, Cognitive Science, Simulator models, AI-assisted modeling
F5: Virtual Laboratories: Drug Design
We develop modeling methods for drug design, both generative models of the drug molecules and their effects, and collaborative AI methods for assisting the drug designers in their task. The idea is to help experts steer the modeling system towards their design goals, while eliciting their prior knowledge to improve the models of the drugs. This is difficult because the goals may be tacit, uncertain and evolving.

We work with leading pharma companies and academic groups in Europe, USA, and Canada. Key methods we will need: probabilistic modeling and Bayesian inference, multi-agent modeling, sequential experimental design, POMDPs, reinforcement learning and inverse reinforcement learning. We expect applicants to master some of these, or be exceptionally eager and quick learners.

Supervision: Prof. Vikas Garg (Aalto University), Prof. Samuel Kaski (Aalto University), Markus Heinonen (Aalto University)

Keywords: Drug design, generative modeling, human-in-the loop machine learning
Topic F6: AI-Assisted Design 
FCAI is working on a new paradigm of AI-assisted design that aims to cooperate with designers by supporting and leveraging the creativity and problem-solving of designers. The challenge for such AI is how to infer designers’ goals and then help them without being needlessly disruptive. We use generative user models to reason about designers’ goals, reasoning, and capabilities. In this call, FCAI is looking for a postdoctoral scholar or research fellow to join our effort to develop AI-assisted design. Suitable backgrounds include deep reinforcement learning, Bayesian inference, cooperative AI, computational cognitive modeling, and user modeling.

Example publications by the team





Supervision: Profs. Antti Oulasvirta (Aalto University), Samuel Kaski (Aalto University), Perttu Hämäläinen (Aalto University)

Keywords: AI-assisted design, user modeling, cooperative AI
F7: AI-Assisted Modeling
We are working on the development of a platform for integrated and semi-automated Bayesian Workflow. We aim at improving the probabilistic programming experience and the outcomes of Bayesian modeling by providing assistance at different steps of model building, while including the modeller in the loop. This will redound in a more pleasant modeling experience, one that is also less prone to errors and bad practices, while helping users to include more domain-knowledge into the modeling process.

Our main expertise is in Bayesian machine learning, and we collaborate actively with other global leaders including DTU, Cambridge, and Columbia in NYC. We also contribute to open source software, including Stan, ArviZ and other libraries. In particular, we focus on 1) The development of high-quality verified software for probabilistic programming. 2) The development of computationally efficient and accurate model-agnostic Bayesian inference algorithms. 3) The theory and tools for understanding the modeling workflow as a whole, integrating model specification, validation, refinement, and visualization.

Examples of publications:

  • Kallioinen, N., Paananen, T., Bürkner, P.-C, Vehtari, A. (2021). Detecting and diagnosing prior and likelihood sensitivity with power-scaling. [preprint](
  • Säilynoja, T., Bürkner, P.-C, Vehtari, A. (2021). Graphical Test for Discrete Uniformity and its Applications in Goodness of Fit Evaluation and Multiple Sample Comparison. [preprint](
  • Dhaka, A. K., Catalina, A., Welandawe, M., Andersen, M. R., Huggins, J., & Vehtari, A. (2021). Challenges and Opportunities in High Dimensional Variational Inference. Advances in Neural Information Processing Systems, 34 [].
  • Gelman, A., Vehtari, A., Simpson, D., Margossian, C. C., Carpenter, B., Yao, Y., Kennedy, L., Gabry, J., Bürkner, P.-C. & Modrák, M. (2020). Bayesian workflow. [preprint](

Supervision: Aki Vehtari (Aalto University), Arto Klami (University of Helsinki).

Keywords: Probabilistic modeling, Bayesian inference, Bayesian workflow
F8: Collaborative AI for AI-Assisted Decision Making
We develop probabilistic modeling and inference techniques that take into account the down-the-line decision making task. A particularly interesting case is delayed-reward decision making where data has to be measured, at a cost, before making the decision. This problem occurs in designing the design-build-test-learn cycles which are ubiquitous in engineering systems, and experimental design in sciences and medicine. The solutions need Bayesian experimental design techniques able to work well with both simulators, measurement data and humans in the loop, who are both information sources and the final decision makers. Furthermore, we need automatic design of interventions (actions) for learning causal models partially from a combination of observational and interventional data.

We are looking for a probabilistic modeling researcher interested in developing the new methods, with options on applying the techniques to improve modeling in the FCAI’s Virtual Laboratories (see Topics 0-5).

Supervision: Profs. Samuel Kaski (Aalto University), Luigi Acerbi (University of Helsinki), other professors involved in the topic

Keywords: Sequential design of experiments, Bayesian experimental design, active learning
F9: Machine learning for collaborative AI
We study how to build collaborative agents (the AI) which are able to help another agent (the user) perform a task.

A prime goal for FCAI’s research is to develop a new form of AI that can better work with people and assist them in everyday tasks. This can be seen as a probabilistic modeling task which requires data-efficient inference on multi-agent models, and some prior knowledge from cognitive science. We are now looking for an outstanding machine learning researcher who wants to develop with us the theory and inference methods for this new task. This will involve multi-agent modeling, POMDPs and reinforcement learning, and inverse reinforcement learning.

A sample previous paper: Tomi Peltola, Mustafa Mert Çelikok, Pedram Daee, Samuel Kaski (2019). Machine Teaching of Active Sequential Learners Conference on Neural Information Processing Systems, NeurIPS 2019

Supervision: Prof. Samuel Kaski (Aalto University), collaborators in Alan Turing Institute, TU Delft, Prof. Antti Oulasvirta (Aalto University)

Keywords: Collaborative AI, inverse reinforcement learning, reinforcement learning, computational cognitive modeling, interactive AI, Multi-agent modeling
F10: ELFI: Engine for Likelihood-free Inference
ELFI ( is a leading software platform for likelihood-free inference of interpretable simulator-based models. The inference engine is built in a modular fashion and contains the most popular likelihood-free inference paradigms, such as ABC and synthetic likelihood, but also more recent approaches based on classifiers and GP emulation for accelerated inference. We are looking for postdoctoral researchers and research fellows to spearhead development of the next-generation version of the inference engine supporting new inference methods, including the use of PyTorch and deep neural networks for amortized inference, and using ELFI in cutting-edge applications from multiple fields of science.

Supervision: Profs. Jukka Corander (University of Helsinki), Luigi Acerbi (University of Helsinki)

Keywords: Machine learning, emulators, simulator-based inference
F11: Design of Maximally Autonomous Collaborative AI Systems
Designing AI-agents that perform sequential tasks for users is challenging, especially in cases where the underlying goal is difficult to specify. The more automatically the AI system can operate, the more it can help us – but if it has not understood what we want, we would not want the help. The problem is compounded when the user is unable to provide ideal demonstrations to the agent due to some constraints. Such scenarios arise in many robotic applications, where providing optimal demonstrations is not straightforward.

We develop methods that can infer the underlying goal of a task through minimal user interaction and feedback. This requires the use of Bayesian experimental design techniques in combination with inference methods and interactive learning.

We are looking for people with a strong background in probabilistic machine learning, in particular reinforcement learning. Prior experience on robotics applications is a plus though not necessary – the principles are broadly applicable beyond robotics.

Supervision: Profs. Ville Kyrki (Aalto University), Samuel Kaski (Aalto University)

Keywords: Probabilistic machine learning, reinforcement learning, Bayesian experimental design, user modeling
F12: Societal Aspects of AI
The development of AI and machine learning has created new opportunities to utilize data, but good governance requires a common understanding of data and methods as well as a grasp of the related complex socio-legal-technical issues. The goal of this project is to increase a holistic understanding of how data are generated, processed, analyzed, and presented, looking at it from the point of view of modern AI methodology, thereby providing a foundation for common understanding about the limitations and opportunities. This improved understanding will lead to better decision-making, more realistic expectations about the possibilities of data analytics, and sharper critical discourse on future dangers.

The project is a close collaboration between social scientists, legal scholars, cognitive and computer scientists, as well as central Finnish institutions that manage much of the data about the Finnish population. The goal of the project is to advance socially and ethically acceptable uses of social and health data in public decision-making.

An ideal candidate has a PhD in machine learning or a related technical field, and a strong passion for interdisciplinary research on the burning societal issues. The role of the candidate is to contribute to both machine learning research with top experts in the field as well as interdisciplinary research together with our strong group of collaborators.

Supervision: Profs. Pekka Marttinen (Aalto University), Petri Ylikoski (University of Helsinki)

Keywords: Uncertainty, bias, fairness, decision making, hidden assumptions, missing data
F13: Machine Learning to Integrate Family Structure into Health Trajectories Across 7.1 Million Individuals
Finland is one of the most advanced countries in the world with regard to collecting and accessing genetic, health and socio-economic data for research purposes. The Finregistry ( and FinnGen ( projects have aggregated an unprecedented amount of genetic, health and socio-demographic information by leveraging the power of nation-wide registers including information on hospitalisation, drug purchases, surgical operations, education, job profession, familial pedigrees, among many others. Our team ( is composed of statisticians, biologists, computer scientists and medical doctors and it is interested in developing and applying statistical and machine learning approaches to combine disease trajectories and genetic information with the goal to improve public health interventions. We are looking for an exceptional post-doctoral candidate that is passionate about understanding why people get sick and how computational approaches can help to identify at-risk individuals. In particular, the candidate will work on an exciting project to integrate health trajectories of family members and relatives for identifying individuals that are at higher risk to develop certain common diseases.

The successful candidate should prove a solid understanding of longitudinal data analysis from a biostatistical and/or machine/deep learning (i.e. recurrent neural networks, transformer models) perspective. Understanding of epidemiological design and measures is considered a plus.

This project will be carried out in collaboration with the Eric and Wendy Schmidt Center ( which provides mobility opportunities between FIMM and the Broad institute and the possibility to directly collaborate with world-leading experts at the intersection of machine learning and health at the Broad Institute, MIT and Harvard community.

Supervision: Andrea Ganna (FIMM, University of Helsinki), co-supervision: Prof. Pekka Marttinen (Aalto University), Anthony Philippakis (Broad Institute)

Keywords: Health data science, machine learning, genetics
F14: Computational Rationality
Computational rationality is an emerging integrative theory of intelligence in humans and machines [1] with applications in human-computer interaction, cooperative AI, and robotics. The  theory assumes that observable human behavior is generated by cognitive mechanisms that are adapted to the structure of not only the environment but also the mind and brain itself [2]. Implementations use deep reinforcement learning to approximate optimal policy within assumptions about cognitive architecture and their bounds. Cooperative AI systems can utilize such models to infer causes behind observable behavior and plan actions and interventions in settings like semiautonomous vehicles, game-level testing, AI-assisted design etc. FCAI researchers are at the forefront in developing computational rationality as a generative model of human behavior in interactive tasks (e.g., [3,4,5]) as well as suitable inference mechanisms [5]. We collaborate with University of Birmingham (Prof. Andrew Howes) and Université Pierre et Marie Curie (UPMC, CNRS) (Dr. Julien Gori, Dr. Gilles Bailly).

In this call, we are looking for a talented postdoctoral scholar or research fellow to join our effort to develop computational theory as a model of human behavior. Suitable backgrounds include deep reinforcement learning, computational cognitive modeling, and reinforcement learning.


[1] S. Gershman et al. Computational rationality: A converging paradigm for intelligence in brains, minds, and machines. Science 2015.

[2] R. Lewis, A. Howes, S. Singh. Computational rationality: A converging paradigm for intelligence in brains, minds, and machines. Topics in Cognitive Science 2014.

[3] J. Jokinen et al. Parameter Inference for Computational Cognitive Models with Approximate Bayesian Computation. Proc. CHI’21, ACM Press.

[4] C. Gebhardt et al. Hierarchical Reinforcement Learning Explains Task Interleaving Behavior. Computational Brain & Behavior 2021.

[5] J. Takatalo et al. Predicting Game Difficulty and Churn Without Players. Proc. CHI Play 2020.

[6] A. Kangasrääsiö et al. Parameter Inference for Computational Cognitive Models with Approximate Bayesian Computation. Cognitive Science 2019.

Supervision: Profs. Antti Oulasvirta (Aalto University), Andrew Howes (University of Birmingham), Samuel Kaski (Aalto University), Arto Klami (University of Helsinki), Perttu Hämäläinen (Aalto University)

Keywords: Computational rationality, computational cognitive modeling, deep reinforcement learning
F15: Privacy-preserving and federated learning
Many applications of machine learning suffer from limited training data availability because data holders cannot share their data. The aim of this project is to develop solutions to this fundamental problem through efficient privacy-preserving learning methods that allow securely combining data from multiple data holders under guarantees that data will not leak. Possible approaches include extending cross-silo federated learning into collaborative learning through personalisation of the models to each party, and developing methods for generating privacy-preserving synthetic data. The security and privacy will be guaranteed by a combination of differential privacy and secure multi-party computation.

In this project, you will join our group in developing new learning methods operating under these guarantees, and applying them to real-world problems. Collaboration opportunities will enable testing the methods on academic and industrial applications. A strong candidate will have a background in machine learning or a related field. Experience in differential privacy and/or secure multi-party computation is an asset.

Supervision: Prof. Antti Honkela, Prof. Samuel Kaski, Prof. Patrick Rinke,

Keywords: Differential privacy, federated learning, personalisation, synthetic data
F16: Reinforcement Learning Under Uncertainty
We are looking for exceptional and highly motivated candidates to work in the interface of model-based reinforcement learning and Bayesian deep learning. One of the main challenges in model-based RL methods is how to choose actions such that you collect information that leads to learning a model that can be used to plan optimal behaviour for a specified task. This project is concerned with solving model-based RL by (i) quantifying the uncertainty by approximative inference methods, and (ii) using the uncertainty estimates to explore states with both high uncertainty and which may be important for solving the task. The goal is to experiment with methods on a Boston Dynamics Spot robot.

Successful candidates are expected to have previous experience in RL and knowledge of probabilistic methods in machine learning. This is a co-located position between Prof. Pajarinen’s Robot Learning Lab ( and Prof. Solin’s ML group ( See group pages for recent publications on the topics.

Supervision: Profs. Joni Pajarinen (Aalto University) and Arno Solin (Aalto University)

Keywords: Reinforcement learning, model-based RL, uncertainty quantification, Bayesian deep learning

How to Apply:

The applications are to be submitted through the eRecruitment system. Choose in the application form one or more of the research centers, programs and/or projects described above and explain in the motivation letter how you could contribute in the selected research area(s).

Required attachments:

(1) Cover letter which includes your motivation for applying for the specific position(s), how you could contribute to them, and other relevant information (no more than 2 pages).

(2) CV

(3) List of publications (please do not attach full copies of publications)

(3) A transcript of doctoral studies and the degree certificate of the doctoral degree

All material should be submitted in English. The application materials will not be returned. Short-listed candidates may be invited for an interview either at the Otaniemi or Kumpula campus or for an interview conducted via Skype.

Apply Through our eRecruitment System Here


The candidate should have a PhD and is expected to have an excellent track record in scientific research in one or more fields relevant to the position. Good command of English is a necessary prerequisite. In the review process, particular emphasis is put on the quality of the candidate’s previous research and international experience, together with the substance, innovativeness, and feasibility of the research plan, and its relevance to the research group or groups in question. Efficient and successful completion of studies is considered an additional merit.

Compensation, working hours and place of work:

The salary for a postdoctoral researcher starts typically from about 3500 EUR per month depending on experience and qualifications. In addition to the salary, the contract includes occupational health benefits, and Finland has a comprehensive social security system. The annual total workload of teaching staff at the recruiting universities is 1624 hours.

The position is located at Aalto University’s Otaniemi campus or the University of Helsinki’s Kumpula campus. The positions belong to the university career system and the selected persons will be appointed for fixed-term positions, for postdoctoral researchers typically for two years (unless otherwise stated) with an option for renewal. For exceptional candidates, a longer term Research Fellow position can be considered. The length of the contract and starting and ending dates are negotiable. In addition to research work, the persons hired are expected to participate in the supervision of students and teaching following the standard practices of the hiring department.

About HIIT and FCAI:

Helsinki Institute for Information Technology HIIT is a joint research institute of Aalto University and the University of Helsinki for basic and applied research on information technology. HIIT’s mission is to conduct top-level research, seamlessly moving between fundamental methods and technologies to novel applications and their impact on people and society. For more information, see

Finnish Center for Artificial Intelligence FCAI brings together the world-class expertise of Aalto University and the University of Helsinki in AI research, strengthened further with an extensive set of companies and public sector partners. FCAI has been selected as one of the prestigious Flagships of the Academy of Finland, a status granted to very few selected centers of excellence with high societal impact. The total budget of FCAI is 250 M€ over the next 8 years. For more information see

Further information:

Please contact the leader leader of the research center/program/project specified above for project specific questions, at  ( or (

Application process and practicalities: Sanni Kirmanen (

For more general information please contact the HIIT coordinator at