Projects

De moord op Don Tomasso

Trailer — Full film — Personal IMDb profile

I directed and produced a feature-length movie which had a limited release in cinemas in the Netherlands and a global release on YouTube.

Synopsis: 12 members of a mafia family travel to their yearly get-together, where even their beloved Don might make an appearance. Soon, they discover there are extra security measures because the family expects one of the 12 members to have a sinister plan to kill Don Tomasso. By performing challenges, the members of the family can find out who the traitor in their midst is, in order to foil the plan to murder the Don.

bnd-quotes

This project delivers quotes from BrandNewDay.nl funds in a structured manner, with support for PortfolioPerformance. It currently serves roughly 10 million requests per month.

National Cyber Security Center (NCSC-NL)

Greenfield data platform project.

Politie

I led a 7-man team building the Advanced Analytics Platform (AAP), a machine learning platform for all analysts and data scientists in the organization. We focus on both model training and model productionization and automate as much as possible for our users. The platform is based on Kubeflow and runs on top of an on-premise Kubernetes cluster. We use MinIO to provide an S3 layer and use Istio as our service mesh. Everything is deployed using Infrastructure-as-Code (Terraform) and modern DevOps practices (ArgoCD, GitOps). As tech lead, I’m responsible for the platform’s architecture, performance, usability and security.

SKF

I implemented the cloud infrastructure of the Diagnostics & Prognostics team on AWS. The team is responsible for predicting failures of the ball bearings SKF produces. I worked on setting up cloud resources using Terraform, created automated deployment pipelines for production-ready applications and built DAGs to run on Airflow. I also deployed Databricks and connected it to AWS Glue so that researchers are able to access all data instantly. I gave a talk at the NL Big Data Expo 2019 about the project.

ING

As Lead of the Productionization Team in the 1:1-Analytics Tribe I acted as the liaison between data scientists and the platform team. I was in the lead for shaping the road to production by creating tools data scientists can use to effortlessly industrialize their models and by educating everyone involved. This involved setting up a Hadoop/Spark cluster using Ansible, developing a continuous deployment pipeline for Docker containers in Jenkins and assisting data scientists in making their applications production-ready. I helped both the data scientists and platform engineers in building their education plans regarding relevant data science and engineering skills and have given them multiple courses on e.g. Spark, Kubernetes and data science productionization.

Nutreco

As a Data Scientist at Nutreco I was involved in setting up the first Big Data Team within the organization. I explored which data sets and business processes were active in the company and created new projects out of my findings. One of these projects involved semantic segmentation of microscopic images to help in QA, for which I developed a highly successful deep learning solution. I also introduced the organization to AWS and Azure and deployed several data pipelines, connecting cloud and on-premise services.

European Space Agency

I did a technical internship at the European Space Agency’s Test Centre. I worked at the Mechanical Engineering Department as an expert in computer vision and machine learning. I created a method for performing a spatial-spectral calibration of the artificial solar beam using a hyperspectral camera. I also built a classification system for space material detection using similar hyperspectral processing techniques. I applied my work in practice for the testing of the BepiColombo spacecraft, which was successfully launched in October 2018 and is expected to insert into Mercury orbit in 2025.