Senior machine learning engineer

5/2019 – 3/2021

Scope: Find optimal parameters for running a heat grid, given the weather conditions and the
structure of the grid.
Approach: Demand forecasting, physical modeling of the grid, modelling the heat losses and
defining an objective function. Optimizing the objective function given the constraints and the
predicted demand.
Results: Working real time system deployed in AWS. The system controls the primary and
secondary part of a heat delivery network of a mid size DSO in the Netherlands as of January 2020.

Amazon Web Services (AWS), Docker, Kubernetes, Python, TensorFlow

3/2019 – 5/2019

Scope: Given load profile data for transformers, identify relevant patterns that can help identify
industrial and residential usage as well as the type of the industrial usage and production.
Results: Successfully distinguished between residential and industrial load patterns. Created an
approach to differentiate between wind power generation, solar generation and gas generators.

Good clinical practice (GCP), Pandas DataFrame, Python

5/2018 – 5/2019

Scope: Extend the RNN model to provide uncertainty of the estimation and work in a Kalman-Filter
like cycle so it can be used with missing data. Evaluation done in the context of pedestrian tracking.
Results: A novel approach for training RNNs with missing data based on the uncertainty of the
estimation and the new proposed RNN-Cycle. Results to be published in paper.

Rekurrentes Neuronales Netzwerk (RNN), TensorFlow, Python