Teemu Säilynoja

Experience

Doctoral researcher2019-2024
Aalto University, Department of Computer Science, Probabilistic Machine Learning Group
Developing principled methods to aid in the Bayesian workflow of probabilistic modelling.
  • Methods for calibration assessment of the model predictions, and the inference algorithms.
  • Recommendations and tools for visual prior and posterior predictive checking.
  • Contributions to open source R packages for probabilistic programming and modelling.
Data scientist2018-2019
Thirdpresence Oy
Machine learning engineering and data analysis on a platform with over 5 billion daily transactions.
  • Developing machine learning models with Python.
  • Serving a scalable prediction API of these models on AWS.
  • Actionable trading data analysis to support the Ad Operations team.

Publications

Posterior SBC: simulation-based calibration checking conditional on dataFebruary 5th, 2025
Preprint
Teemu Säilynoja, Marvin Schmitt, Paul Bürkner, and Aki Vehtari
arXiv:2502.03279

Graphical test for discrete uniformity and its applications in goodness-of-fit evaluationMarch 24th, 2022
and multiple sample comparison
Statistics and Computing 32, no. 2
Teemu Säilynoja, Paul-Christian Bürkner, and Aki Vehtari
doi:10.1007/s11222-022-10090-6

Education

Master of Science - Mathematical analysis2016 - 2017
University of Helsinki, Department of Mathematics and Statistics
Master’s Thesis: Hamilton-Jacobi Equations and Scalar Conservation Laws

Grade: Eximia Cum Laude Approbatur

Bachelor of Science - Mathematics2011-2016
University of Helsinki, Department of Mathematics and Statistics
Bachelor’s Thesis: Method of Two-Block Partial Least-Squares in Morphology

Grade: 5/5

Skills

Strengths

Problem solving, analytical thinking, scientific writing, communication of complex topics.

Programming

R, Python, Stan, Rust, (SQL, Java, C++, Matlab)

Data Science

R (brms, tidyverse), Python (PyTorch, Scikit-learn, Numpy, Pandas), Stan

Data Visualisation

R (Ggplot2, Shiny, Bayesplot), Python (Matplotlib, Seaborn, Shiny)

Languages

Finnish (native), English (fluent, bilingual), Swedish (basic conversation skills)

Professional development

Nordic Probabilistic AI School 20232023
Norwegian University of Science and Technology, Trondheim

A hands-on intensive course on state-of-the-art machine learning and artificial intelligence techniques. The course included hands-on tutorials and a course project of implementing a diffusion probabilistic model, a generative AI model combining neural networks and probabilistic modelling.

Teaching

Stancon 2024 - Model selection tutorial2024
Oxford University, United Kingdom
Model selection, model validation, hypothesis testing, and model comparison in probabilistic modelling.
  • Together with my colleagues, we organised a tutorial on the above topics.
  • The participants brought their own datasets and we guided them in implementing the methods.
Stan Connect 2021 - SBC talk & tutorial2021
Online
  • I gave a talk introducing research, and how it can be applied in simulation-based calibration checking (SBC).
  • I was also part of the three-person team which organised and led a tutorial on implementing SBC as part of the probabilistic machine learning workflow.
Bayesian Data Analysis - Teaching Assistant2019-2024
Aalto University, Department of Computer Science
An annual course of 300 students, providing an in-depth overview of Bayesian data analysis and probabilistic modelling.
  • Teaching bayesian statistics, prior selection, posterior predictive checking, model comparison, model validation, convergence diagnostics, etc.
  • Grading the course projects and presentations of the students.
Course Assistant2012-2017
University of Helsinki, Department of Mathematics and Statistics
Several undergraduate level courses including courses on calculus, ordinary differential equations, basics of propositional and predicate logic, and model theory.
  • Holding weekly exercise sessions.
  • Writing the solution examples for the exercises.
  • Grading course exams, and assisting in designing questions for the exercises and some of the exams.
  • Most courses employed the method of extreme apprenticeship, focusing teaching on the problem solving phase of the exercises.
Teacher & Teacher Assistant2012-2013, 2015
University of Helsinki, Department of Mathematics and Statistics
I was the head teacher, and prior to that the assistant teacher, on a course revising topics of the Finnish High School Mathematics Curriculum.
  • Planning and giving weekly on demand lectures and personal counselling.
Additionally, I worked as an Assistant Teacher on the course titled Introduction to Teaching Mathematics
  • Preparing and giving some of the lectures and providing the study counselling for the students.