Several Openings for Postdocs and PhD students in Machine Learning

Finnish Center for Artificial Intelligence FCAI and ELLIS Unit Helsinki invite applications for research positions in machine learning. You will join one of the top AI research centers in the Nordics and in Europe, with access to an excellent network of scientists and a broad range of possibilities to work with companies.

Jump to:

Areas of research
Who we are
What we are looking for
Our offer
How to apply
FAQ

We are looking for postdocs and PhD students to FCAI and ELLIS Unit Helsinki. Your research can be theoretical, applied, or both. The positions are in the following areas of research:

1) Reinforcement learning
2) Probabilistic methods
3) Simulation-based inference
4) Privacy-preserving machine learning
5) Collaborative AI and human modeling
6) Machine learning for science

You will join a community of machine learning researchers and will be part of a broader team of researchers studying similar topics, mentored by a group of several experienced professors. 

Areas of research

1) Reinforcement learning

We develop reinforcement learning techniques to enable interaction across multiple agents including AIs and humans. We also work on manifold applications, ranging from drug design to autonomous traffic. Examples of recent research include:

• Generating Code World Models with Large Language Models Guided by Monte Carlo Tree Search (NeurIPS 2024)

• Probabilistic Subgoal Representations for Hierarchical Reinforcement learning (ICML 2024)

• Beyond Uniform Sampling: Offline Reinforcement Learning with Imbalanced Datasets (NeurIPS 2023)

• Best-Response Bayesian Reinforcement Learning with Bayes-adaptive POMDPs for Centaurs (AAMAS 2022)

• Training and evaluation of deep policies using reinforcement learning and generative models (JMLR 2022)

• Precise atom manipulation through deep reinforcement learning (Nat. Comms. 2022)

2) PROBABILISTIC METHODS

We develop AI tools using probabilistic programming. Our expertise includes Bayesian machine learning, generative modeling (e.g., diffusion models and GANs in general generative AI) and other probabilistic modeling. The research is disseminated as modular open-source software, including software for the most popular probabilistic programming framework Stan. Examples of recent research include:

Generative modeling

• Alignment is Key for Applying Diffusion Models to Retrosynthesis (ICML 2024)

• Compositional PAC-Bayes: Generalization of GNNs with persistence and beyond (NeurIPS 2024)

• Improving robustness to corruptions with multiplicative weight perturbations (NeurIPS 2024, Spotlight Paper)

• Kernel Language Entropy: Fine-grained Uncertainty Quantification for LLMs from Semantic Similarities (NeurIPS 2024)

• Compositional sculpting of iterative generative processes (NeurIPS 2023)

• Generative modelling with inverse heat dissipation (ICLR 2023)

• Practical Equivariances via Relational Conditional Neural Processes (NeurIPS 2023)

 Active learning and experimental design

• Bayesian Active Learning in the Presence of Nuisance Parameters (UAI 2024)

• Memory-based dual Gaussian processes for sequential learning (ICML 2023)

Other probabilistic modeling

• Detecting and diagnosing prior and likelihood sensitivity (Statistics and Computing 2024) + priorsense package (software)

• Learning high-dimensional mixed models via amortized variational inference (ICML 2024)

• Practical Hilbert space approximate Bayesian Gaussian processes for probabilistic programming (Statistics and Computing 2023)

• Prior knowledge elicitation: The past, present, and future (Bayesian Analysis 2023)

3) Simulation-based inference

We develop simulation-based methods to learn generative models from the data. Main initiatives include: (1) ELFI, a leading software platform for likelihood-free inference of interpretable simulator-based models and (2) numerous leading GAN-based technologies. Examples of recent results:

• Guiding a Diffusion Model with a Bad Version of Itself (NeurIPS 2024, Runner Up for Best Paper Award)

• Learning Robust Statistics for Simulation-based Inference under Model Misspecification (NeurIPS 2023)

• Visualization of extensive datasets (Statistics and Computing 2023; Phil. Trans 2022)

• ABC of the future (International Statistical Review 2022)

• Causal discovery for the microbiome (Lancet Microbe 2022)

• Alias-Free Generative Adversarial Networks (NeurIPS 2021) + StyleGANs (software)

• Cost-aware Simulation-based Inference (preprint)

4) PRIVACY-preserving Machine LEARNING

We develop theory and methods for privacy-preserving machine learning using differential privacy. We focus especially on high-utility differentially private deep learning and differentially private synthetic data. Examples of recent research include: 

• Noise-Aware Differentially Private Regression via Meta-Learning (NeurIPS 2024)

• Subsampling is not Magic: Why Large Batch Sizes Work for Differentially Private Stochastic Optimisation (ICML 2024)

• Towards Efficient and Scalable Training of Differentially Private Deep Learning (ICML 2024)

• Individual Privacy Accounting with Gaussian Differential Privacy (ICLR 2023)

• Noise-Aware Statistical Inference with Differentially Private Synthetic Data (AISTATS/PMLR 2023)

• On the Efficacy of Differentially Private Few-shot Image Classification (TMLR 2023)

5) Collaborative AI and human modeling

We develop probabilistic methods and inference techniques for reinforcement and machine learning in assistance settings with realistic and interactive user models. These systems treat and model human users as active agents to reason over and collaborate with, instead of passive sources of data. Examples of recent research include: 

• Attaining Human’s Desirable Outcomes in Human-AI Interaction via Structural Causal Games (ICML 2024)

• CRTypist: Simulating Touchscreen Typing Behavior via Computational Rationality (CHI 2024, honorable mention)

• Open Ad Hoc Teamwork with Cooperative Game Theory (ICML 2024)

• Preference Learning of Latent Decision Utilities with a Human-like Model of Preferential Choice (NeurIPS 2024)

• Supporting Task Switching with Reinforcement Learning  (CHI 2024, honorable mention)

• AI-assisted design with human-in-the-loop (AI Magazine 2023)

• Amortized inference with user simulations (ACM CHI 2023)

• Differentiable user models (UAI 2023)

• Multi-fidelity bayesian optimization with unreliable information sources (AISTATS 2023)

• Towards machines that understand people (AI Magazine 2023)

• Zero-shot assistance in sequential decision problems (AAAI 2023)

6) Machine learning for science

Machine learning is increasingly being used as a key element in research in different fields. Our interest lies in the general question of how machine learning could be used as part of the research process, essentially to improve the results and the scientific process itself. We seek solutions that work across multiple disciplines and applications. The work relates closely to our Virtual Laboratories initiative. Examples of recent initiatives include:

• Virtual Laboratories: Transforming research with AI (perspective piece, Data-Centric Engineering 2024)

• Modular pipeline for design assistance (software platform (WIP))

• Amortized Bayesian Experimental Design for Decision-Making (NeurIPS 2024)

• Diffusion Twigs with Loop Guidance for Conditional Graph Generation (NeurIPS 2024)

• Recursive Nested Filtering for Efficient Amortized Bayesian Experimental Design (NeurIPS 2024)

• Virtual Laboratory for Molecular Level Atmospheric Transformations Centre of Excellence; publication example (virtual laboratory, center of excellence)

• Engineering new enzymes with machine learning (research project)

What we are looking for

You have previous experience in machine learning, statistics, artificial intelligence or a related field, preferably demonstrated by success in related studies (PhD student applicants) and/or publication record in the leading machine learning venues, e.g. AAAI, AISTATS, ICLR, ICML, JMLR, NeurIPS, (postdoc/research fellow applicants). Other merits demonstrating suitability for a researcher position can also be considered.

You hold (or expect to shortly receive) a Master’s degree (PhD student applicants) or a PhD (postdoc applicants) in computer science, statistics, electrical engineering, mathematics or a related field.

Experienced postdoc applicants can be considered for research fellow positions, typically having previously worked successfully as postdocs for several years.

The positions require the ability to work both independently and as part of a team in a highly collaborative and interdisciplinary environment.

Our offer

1) Research environment and supervision

FCAI’s research mission is to create new types of AI that are data-efficient, trustworthy, and understandable. We work towards this by developing machine learning principles and methods, and by building AI systems capable of helping their users make better decisions and design sustainable solutions across a range of tasks from health applications to materials science. Examples of latest research results are highlighted, e.g., in NeurIPS 2024 conference.

You will join a community of machine learning researchers who all make important contributions to our common agenda, providing each other new ideas, complementary methods, and attractive case studies. Your research can be theoretical, applied, or both.

In this call, we are primarily recruiting new researchers to FCAI Teams, groups of postdocs and PhD students studying similar topics, mentored by a group of several experienced professors. Recruited researchers will typically be supervised by two professors in FCAI and will join two FCAI Teams that support their academic work. Additionally, several supervisors in FCAI are also looking for postdocs and PhD students to join their own research projects; you can express your interest to also apply to the positions of individual research groups in your application.

Read more about the FCAI Teams →
See the list of supervisors →

Our research environment provides you with a broad range of possibilities to work with companies and academic partners, and supports your growth as a researcher. FCAI, host of ELLIS Unit Helsinki, is a salient part of the pan-European ELLIS network, which further strengthens our collaboration with other leading machine learning researchers in Europe. In 2025, these already significant activities will experience a major step forward, as ELLIS Institute Finland will be launched bringing a significant amount of new experts and investments into AI research in Finland.

In addition, our local and national computational services give our researchers access to excellent computing facilities, spearheaded by the EuroHPC LUMI, one of the the fastest and greenest supercomputers in the world. With a new pan-European supercomputer and AI Factory, the computing infrastructure will be significantly developed in the coming years. FCAI will be strongly involved in this upskilling, knowledge transfer, and leveraging AI-optimized supercomputing capabilities.

2) Job details

The positions are based either at Aalto University or at the University of Helsinki, depending on the primary supervisor. All positions are fully funded, and the salaries are based on the Finnish universities’ pay scale. The contract includes occupational healthcare. 

Postdoc positions are typically made for up to three years. Following the standard practice, the PhD student position contract will be made initially for two years, then extended to another two years after a successful mid-term progress review.

Starting dates are flexible. All positions are negotiated on an individual basis and may include, e.g., a relocation bonus, an independent travel budget or research software engineering support.

We are strongly committed to offering everyone an inclusive and non-discriminating working environment. We warmly welcome qualified candidates from all backgrounds to apply and particularly encourage applications from women and other groups underrepresented in the field. Our community is fully international, and the working language is English.

Who we are

Finnish Center for Artificial Intelligence FCAI is an international research hub initiated by Aalto University, the University of Helsinki, and the Technical Research Centre of Finland VTT. We are part of ELLIS, the pan-European AI network of excellence, and we host ELLIS Unit Helsinki.

FCAI and ELLIS Unit Helsinki are built on a long track record of pioneering machine learning research. We create methods and tools for AI-assisted decision-making, design and modeling, and use them to renew industry and society. Currently, over 70 professors contribute to our research. (See the list of supervisors in this call here.)

Our researchers have access to excellent computing facilities through local and national computational services, spearheaded by the EuroHPC supercomputer LUMI, one of the fastest supercomputers in the world.

Our community organizes frequent seminars, e.g., ELLIS Distinguished Lectures, Machine Learning Coffee Seminar and Seminar on Advances in Probabilistic Machine Learning. We offer high-quality collaboration opportunities with other leading research networks and companies. For instance, FCAI has a joint research center with NVIDIA and Finnish IT Centre for Science CSC and collaborates closely with the Alan Turing Institute.

About Finland

Finland is a great place for living with or without family: it is a safe, politically stable, and well-organized Nordic society, where equality is highly valued and extensive social security supports people in all life situations.

Finland's free high-quality education system is also internationally renowned. Finland has been listed as the happiest country in the world for the seventh year running. Find more information about living in Finland here and here.

Finland’s universities are committed to promoting equality and inclusion and preventing discrimination, to ensure that all students and staff feel welcome at universities and that it is easy to come and study or work in Finland.

Stories of people at FCAI

See more stories →

How to apply?

The call opens on December 18 and closes on February 2, 2025. You can apply through the online application system linked below. Please note that there are separate applications for the postdoctoral and PhD student positions.

Required attachments:

1. Cover letter (1–2 pages). Please specify the FCAI Teams and supervisors with whom you want to work.

2. CV

3. List of publications (if applicable). Please highlight the most relevant publications. Do not attach full copies of publications.

4. A transcript of doctoral, MSc and BSc studies and the degree certificate of your latest degree. If you are applying for a postdoc position and don’t yet have a PhD degree or for a PhD student position and don’t have a Master's degree, a plan of completion must be submitted.

5. In the application form, you are also asked to specify contact details of two senior academics who can provide references. We will contact your referees if we need recommendation letters.

All materials should be submitted in English in a PDF format. Note: You can upload max. five files to the recruitment system, each max. 5MB.

Frequently asked questions