Jan Thiemen Postema

MSC

Consultant

Amsterdam, NL

Jan Thiemen Postema works as a consultant in the Amsterdam office of Milliman.

Loading component...

Experience

Jan Thiemen joined Milliman in 2020, and has five years of experience working in the realm of data science. Before joining Milliman, he worked in a similar role at Ortec Finance. At Milliman, he focusses mainly on the application of data science and machine learning techniques in the insurance sector, as well as data engineering and visualization projects.

Education

Master of Science in Computer Science, specialization Data Science, University of Twente, the Netherlands
Bachelor of Science in Business & IT, University of Twente, the Netherlands

Publications

Read their latest work

Article

Where is Europe’s next flash flood? Quantifying extreme precipitation and flash flood risk in Europe with extreme value theory

18 December 2025 - by Matt Chamberlain, Fernando Mierzejewski, Jan Thiemen Postema, Thomas van der Valk, Menno van Wijk

We use extreme value theory to assess the risk of catastrophic precipitation events in Europe, identifying the highest risks along the Mediterranean coast.

Article

Accidents that never happened: Generative AI and fraud in motor insurance

03 December 2025 - by Bartosz Gaweda, Raymond van Es, Leonid Zeldin, Jan Thiemen Postema, Tristan Stoevelaar

As fraudsters submit AI-manipulated images of car crashes, Milliman’s research is helping global motor insurers separate the real claims from the scams.

Article

Clustering techniques: Innovations and practical implementation

29 October 2025 - by Aatman Dattani, Avi Freedman, Corey Grigg, Jan Thiemen Postema, Karol Maciejewski, Sijbo Holtman

We explore evolving model compression techniques that can help insurers achieve significant computational efficiency gains.

Article

Flood risk modelling in Europe

02 May 2024 - by Jan Thiemen Postema, Daniël van Dam, Menno van Wijk, Niels Van der Laan, Antoine Rainaud, Eve Titon

We present a framework and tools for projecting insured flood losses in the Netherlands and France, which can be used for insurance portfolios across Europe.

Article

Impact of COVID-19 on best estimate mortality assumptions

08 September 2023 - by Flora Auter, Amal Elfassihi, Salima el Khababi, Jan Thiemen Postema, Eve Titon, Raymond van Es

Using mortality data adjustment methods for COVID-19 helps avoid double counting and derives long-term trends without instabilities from a one-off experience.

Article

FICO® Score 10 and FICO® Score 10 T model assessment

25 July 2023 - by Kenneth A. Bjurstrom, Raymond van Es, Ryan Huff, Jan Thiemen Postema

FICO 10 and FICO 10 T mark a meaningful advancement in assessing credit in mortgage lending, powered by trended data and cutting-edge analytical techniques.

Article

Anomaly detection techniques in fraud detection, performance optimization, and data quality

08 March 2023 - by Bjorn Blom, Jan Thiemen Postema, Rens IJsendijk, Judith Houtepen, Job Prince

Methods to detect anomalies can be used to find fraudulent claims in insurance, especially in products with a large frequency of payments, such as in healthcare.

Article

Explainable AI in fraud detection

22 December 2022 - by Inès Zitouni, Jan Thiemen Postema, Dominik Sznajder, Raymond van Es

Techniques in explainable artificial intelligence can indicate the reasons why a model expects a claim to be truly fraudulent, saving time for investigators.

Article

Improved mortality rate forecasting using machine learning and open data

23 November 2022 - by Jan Thiemen Postema, Raymond van Es

Compared with traditional techniques, state-of-the-art machine learning algorithms can substantially improve the forecasts of mortality rates.

Article

Developing fair models in an imperfect world: How to deal with bias in AI

23 December 2021 - by Daniël van Dam, Raymond van Es, Jan Thiemen Postema

We need to start training models that reflect the world we want to live in.

Loading component...

Read their latest work

Article

Where is Europe’s next flash flood? Quantifying extreme precipitation and flash flood risk in Europe with extreme value theory

18 December 2025 - by Matt Chamberlain, Fernando Mierzejewski, Jan Thiemen Postema, Thomas van der Valk, Menno van Wijk

We use extreme value theory to assess the risk of catastrophic precipitation events in Europe, identifying the highest risks along the Mediterranean coast.

Article

Accidents that never happened: Generative AI and fraud in motor insurance

03 December 2025 - by Bartosz Gaweda, Raymond van Es, Leonid Zeldin, Jan Thiemen Postema, Tristan Stoevelaar

As fraudsters submit AI-manipulated images of car crashes, Milliman’s research is helping global motor insurers separate the real claims from the scams.

Article

Clustering techniques: Innovations and practical implementation

29 October 2025 - by Aatman Dattani, Avi Freedman, Corey Grigg, Jan Thiemen Postema, Karol Maciejewski, Sijbo Holtman

We explore evolving model compression techniques that can help insurers achieve significant computational efficiency gains.

Article

Flood risk modelling in Europe

02 May 2024 - by Jan Thiemen Postema, Daniël van Dam, Menno van Wijk, Niels Van der Laan, Antoine Rainaud, Eve Titon

We present a framework and tools for projecting insured flood losses in the Netherlands and France, which can be used for insurance portfolios across Europe.

Article

Impact of COVID-19 on best estimate mortality assumptions

08 September 2023 - by Flora Auter, Amal Elfassihi, Salima el Khababi, Jan Thiemen Postema, Eve Titon, Raymond van Es

Using mortality data adjustment methods for COVID-19 helps avoid double counting and derives long-term trends without instabilities from a one-off experience.

Article

FICO® Score 10 and FICO® Score 10 T model assessment

25 July 2023 - by Kenneth A. Bjurstrom, Raymond van Es, Ryan Huff, Jan Thiemen Postema

FICO 10 and FICO 10 T mark a meaningful advancement in assessing credit in mortgage lending, powered by trended data and cutting-edge analytical techniques.

Article

Anomaly detection techniques in fraud detection, performance optimization, and data quality

08 March 2023 - by Bjorn Blom, Jan Thiemen Postema, Rens IJsendijk, Judith Houtepen, Job Prince

Methods to detect anomalies can be used to find fraudulent claims in insurance, especially in products with a large frequency of payments, such as in healthcare.

Article

Explainable AI in fraud detection

22 December 2022 - by Inès Zitouni, Jan Thiemen Postema, Dominik Sznajder, Raymond van Es

Techniques in explainable artificial intelligence can indicate the reasons why a model expects a claim to be truly fraudulent, saving time for investigators.

Article

Improved mortality rate forecasting using machine learning and open data

23 November 2022 - by Jan Thiemen Postema, Raymond van Es

Compared with traditional techniques, state-of-the-art machine learning algorithms can substantially improve the forecasts of mortality rates.

Article

Developing fair models in an imperfect world: How to deal with bias in AI

23 December 2021 - by Daniël van Dam, Raymond van Es, Jan Thiemen Postema

We need to start training models that reflect the world we want to live in.

Jan Thiemen Postema

Loading component...

Experience

Read their latest work

Where is Europe’s next flash flood? Quantifying extreme precipitation and flash flood risk in Europe with extreme value theory

Accidents that never happened: Generative AI and fraud in motor insurance

Clustering techniques: Innovations and practical implementation

Flood risk modelling in Europe

Impact of COVID-19 on best estimate mortality assumptions

FICO® Score 10 and FICO® Score 10 T model assessment

Anomaly detection techniques in fraud detection, performance optimization, and data quality

Explainable AI in fraud detection

Improved mortality rate forecasting using machine learning and open data

Developing fair models in an imperfect world: How to deal with bias in AI

Loading component...

CHOOSE A LOCATION AND LANGUAGE

Jan Thiemen Postema

Read their latest work

Where is Europe’s next flash flood? Quantifying extreme precipitation and flash flood risk in Europe with extreme value theory

Accidents that never happened: Generative AI and fraud in motor insurance

Clustering techniques: Innovations and practical implementation

Flood risk modelling in Europe

Impact of COVID-19 on best estimate mortality assumptions

FICO® Score 10 and FICO® Score 10 T model assessment

Anomaly detection techniques in fraud detection, performance optimization, and data quality

Explainable AI in fraud detection

Improved mortality rate forecasting using machine learning and open data

Developing fair models in an imperfect world: How to deal with bias in AI

CHOOSE A LOCATION AND LANGUAGE