Roel Heremans - Senior Data Scientist

Roel Heremans - Senior Data Scientist

Creating solutions that will change the world.

Creating solutions that will change the world.

Over two decades of expierence working on over 50 projects.

Over two decades of expierence working on over 50 projects.

Download CV

Download CV

Download CV

Accepting New Clients

You're in good company

You're in good company

Bef4U Project

Bef4U, a company specializing in energy efficiency for industry and large tertiary sectors, contacted me to help build a machine learning model to detect electric consumption anomalies. I designed an online approach based on a sliding training and testing window. In each step, an XGBoost model was trained, and a z-score was calculated for anomaly detection, improving energy efficiency monitoring for shopping malls.

Premium UX Template for Framer

Bef4U Project

Bef4U, a company specializing in energy efficiency for industry and large tertiary sectors, contacted me to help build a machine learning model to detect electric consumption anomalies. I designed an online approach based on a sliding training and testing window. In each step, an XGBoost model was trained, and a z-score was calculated for anomaly detection, improving energy efficiency monitoring for shopping malls.

Premium UX Template for Framer

Electrolux Project

For Electrolux, a global leader in household appliances, I developed a regression model using Markov Random Fields to predict the service call ratio (SCR) based on one year of historical data, achieving a 4.2% Mean Absolute Percentage Error (MAPE) for the food preservation category. The project successfully provided precise service call forecasts, enhancing customer service efficiency.

Premium UX Template for Framer

Electrolux Project

For Electrolux, a global leader in household appliances, I developed a regression model using Markov Random Fields to predict the service call ratio (SCR) based on one year of historical data, achieving a 4.2% Mean Absolute Percentage Error (MAPE) for the food preservation category. The project successfully provided precise service call forecasts, enhancing customer service efficiency.

Premium UX Template for Framer

Uptime

The Oxford Uptime team looked for an artificial intelligence solution to detect when water is pumped from wells using offline temperature sensors. I developed a Python-based system comprising a dashboard for visualization and parameter tuning, and a batch processing engine for analyzing sensor data. The AI model learned to identify pump usage by detecting temperature changes when water was pumped. The system enhanced the efficiency of rural water maintenance services.

Premium UX Template for Framer

Uptime

The Oxford Uptime team looked for an artificial intelligence solution to detect when water is pumped from wells using offline temperature sensors. I developed a Python-based system comprising a dashboard for visualization and parameter tuning, and a batch processing engine for analyzing sensor data. The AI model learned to identify pump usage by detecting temperature changes when water was pumped. The system enhanced the efficiency of rural water maintenance services.

Premium UX Template for Framer

Your Political Monitor

Your Political Monitor, led by CEO Caspar Pelle, aimed to create an advanced AI-powered web app for summarizing political messages. I was tasked with developing a demo web app where users could input text or upload PDFs to receive concise summaries without relying on Chat GPT. The app allows users to enter text, generate summaries, and view details on word count and processing time, providing accurate and timely summaries of political content.

Premium UX Template for Framer

Your Political Monitor

Your Political Monitor, led by CEO Caspar Pelle, aimed to create an advanced AI-powered web app for summarizing political messages. I was tasked with developing a demo web app where users could input text or upload PDFs to receive concise summaries without relying on Chat GPT. The app allows users to enter text, generate summaries, and view details on word count and processing time, providing accurate and timely summaries of political content.

Premium UX Template for Framer

fbeta

fbeta offers all-in-one support and expert consulting for healthcare innovators to collaboratively enhance health care for citizens, insured individuals, and patients. They use expert knowledge combined with Large Language Model technologies.

Premium UX Template for Framer

fbeta

fbeta offers all-in-one support and expert consulting for healthcare innovators to collaboratively enhance health care for citizens, insured individuals, and patients. They use expert knowledge combined with Large Language Model technologies.

Premium UX Template for Framer

Bef4U Project

Bef4U, a company specializing in energy efficiency for industry and large tertiary sectors, contacted me to help build a machine learning model to detect electric consumption anomalies. I designed an online approach based on a sliding training and testing window. In each step, an XGBoost model was trained, and a z-score was calculated for anomaly detection, improving energy efficiency monitoring for shopping malls.

Electrolux Project

Electrolux, a global leader in household appliances. I developed a regression model using Markov Random Fields to predict the service call ratio (SCR) based on one year of historical data, achieving a 4.2% Mean Absolute Percentage Error (MAPE) for the food preservation category.The project successfully provided precise service call forecasts, enhancing customer service efficiency.

Magenta Insight LTD

Magenta Insight LTD. in Oxford sought an AI solution to detect when water is pumped from wells using offline temperature sensors. I developed a Python-based system comprising a dashboard for visualization and parameter tuning, and a batch processing engine for analyzing sensor data. The AI model learned to identify pump usage by detecting temperature changes when water was pumped. The system enhanced the efficiency of rural water maintenance services.

Your Political Monitor

Your Political Monitor, led by CEO Caspar Pelle, aimed to create an advanced AI-powered web app for summarizing political messages. I was tasked with developing a demo web app where users could input text or upload PDFs to receive concise summaries without relying on Chat GPT. The app allows users to enter text, generate summaries, and view details on word count and processing time, providing accurate and timely summaries of political content.

fbeta

fbeta offers all-in-one support and expert consulting for healthcare innovators to collaboratively enhance health care for citizens, insured individuals, and patients. They use expert knowledge combined with Large Language Model technologies.

Skills

Skills

  • Regression

    Classification

    Clustering

    Supervised vs. Unsupervised Learning

    NumPy

    Matplotlib

    Train/Test Splits

    Cross-Validation

    Object-Oriented Programming (OOP)

    Data Cleaning and Imputation

    Feature Engineering

    Supervised Learning

  • Pandas

    Seaborn

    Scikit-Learn

    Pytorch

    SQL Basics

    Using SQLite with Python

    Unit Testing with unittest

    Debugging Techniques

    Unsupervised Learning

    Advanced Algorithms

    Cross-Validation

Process

Your compass to innovation and data excellence. With a background in physics and a passion for technology, I remain committed to driving impactful solutions through data-driven insights.

  • 01

    Discover

    In the discovery phase, I collaborate closely with stakeholders to understand their goals and challenges, while gathering and exploring relevant datasets to define the problem statement and key metrics for success.

  • Define

    Through meticulous data cleaning and exploratory analysis, I refine the project scope and select appropriate modeling techniques, ensuring clarity on the problem definition and the methodology to address it effectively.

    02

  • Develop

    During development, I focus on training and refining models iteratively, optimizing their performance and usability while concurrently designing intuitive user interfaces to facilitate understanding and utilization of the insights generated.

    03

  • Deliver

    In the final phase, I ensure the successful deployment of the solution, monitoring its performance over time and providing comprehensive documentation and reporting to stakeholders to support decision-making and ongoing maintenance efforts.

    04

Process

Your compass to innovation and data excellence. With a background in physics and a passion for technology, I remain committed to driving impactful solutions through data-driven insights.

01

Discover

In the discovery phase, I collaborate closely with stakeholders to understand their goals and challenges, while gathering and exploring relevant datasets to define the problem statement and key metrics for success.

Develop

During development, I focus on training and refining models iteratively, optimizing their performance and usability while concurrently designing intuitive user interfaces to facilitate understanding and utilization of the insights generated.

03

Define

Through meticulous data cleaning and exploratory analysis, I refine the project scope and select appropriate modeling techniques, ensuring clarity on the problem definition and the methodology to address it effectively.

02

Deliver

In the final phase, I ensure the successful deployment of the solution, monitoring its performance over time and providing comprehensive documentation and reporting to stakeholders to support decision-making and ongoing maintenance efforts.

04

Process

Your compass to innovation and data excellence. With a background in physics and a passion for technology, I remain committed to driving impactful solutions through data-driven insights.

  • 01

    Discover

    In the discovery phase, I collaborate closely with stakeholders to understand their goals and challenges, while gathering and exploring relevant datasets to define the problem statement and key metrics for success.

  • Define

    Through meticulous data cleaning and exploratory analysis, I refine the project scope and select appropriate modeling techniques, ensuring clarity on the problem definition and the methodology to address it effectively.

    02

  • Develop

    During development, I focus on training and refining models iteratively, optimizing their performance and usability while concurrently designing intuitive user interfaces to facilitate understanding and utilization of the insights generated.

    03

  • Deliver

    In the final phase, I ensure the successful deployment of the solution, monitoring its performance over time and providing comprehensive documentation and reporting to stakeholders to support decision-making and ongoing maintenance efforts.

    04

Recent Certificates

Recent Certificates

Neural Networks and Deep Learning

Neural Networks and Deep Learning

DeepLearning.AI

DeepLearning.AI

Nov 2023

Nov 2023

Structuring Machine Learning Projects

Structuring Machine Learning Projects

DeepLearning.AI

DeepLearning.AI

Dec 2023

Dec 2023

Improving Deep Neural Networks

Improving Deep Neural Networks

DeepLearning.AI

DeepLearning.AI

Dec 2023

Dec 2023

Convolutional Neural Networks

Convolutional Neural Networks

DeepLearning.AI

DeepLearning.AI

Jan 2024

Jan 2024

Sequence Models

Sequence Models

DeepLearning.AI

DeepLearning.AI

Jan 2024

Jan 2024

Generative Pre-trained Transformers

Generative Pre-trained Transformers

University of Glasgow

University of Glasgow

U. of Glasgow

Jan 2024

Jan 2024

Natural Language Processing

Natural Language Processing

DeepLearning.AI

DeepLearning.AI

Feb 2024

Feb 2024

AI with Large Language Models

AI with Large Language Models

DeepLearning.AI

DeepLearning.AI

Feb 2024

Feb 2024

Google AI Essentials

Google AI Essentials

Google Career Certificates

Google Career Certificates

Jun 2024

Jun 2024

Milestone - Experiences

Milestone - Experiences

Background and Education

PhD in Elementary Particle Physics

I began my academic journey as a physicist, earning a PhD in Elementary Particle Physics. During my PhD, I was fascinated by seeing how particles in an accelerator could change their physical properties, like weight and the perception of time. After six years of postdoctoral research, I felt confined to a narrow field of expertise and decided to transition to the industry, seeking broader applications for my skills.

Background and Education

PhD in Elementary Particle Physics

I began my academic journey as a physicist, earning a PhD in Elementary Particle Physics. During my PhD, I was fascinated by seeing how particles in an accelerator could change their physical properties, like weight and the perception of time. After six years of postdoctoral research, I felt confined to a narrow field of expertise and decided to transition to the industry, seeking broader applications for my skills.

Early Career

Career at Royal Military Academy

I spent nine years at the Royal Military Academy in Belgium, working on diverse military-related research projects. My focus was on automating image processing, such as detecting floods on synthetic Aperture radar images and performing 3D image reconstruction using Terra Hertz radiation for body scanners in airports. Also, I enhanced the image quality of synthetic aperture sonar images for mine detection on the sea floor using the DPCA (Displaced Phase Center Array) method and used hyperspectral imaging to determine atomic composition at the pixel level, demonstrating my ability to innovate across different technologies.

Early Career

Career at Royal Military Academy

I spent nine years at the Royal Military Academy in Belgium, working on diverse military-related research projects. My focus was on automating image processing, such as detecting floods on synthetic Aperture radar images and performing 3D image reconstruction using Terra Hertz radiation for body scanners in airports. Also, I enhanced the image quality of synthetic aperture sonar images for mine detection on the sea floor using the DPCA (Displaced Phase Center Array) method and used hyperspectral imaging to determine atomic composition at the pixel level, demonstrating my ability to innovate across different technologies.

Career at AGT International

Sensor-Based Environmental Monitoring

Joining AGT International, I worked on projects involving participatory sensing campaigns using environmental sensors in urban settings. I developed models to temporal heatmaps of the pollution in space and time. Those pollution heatmaps served as inputs for our Healthy route recommendations for city dwellers, which were calculated based on self developed multiple objective shortest path algorithms.

Career at AGT International

Sensor-Based Environmental Monitoring

Joining AGT International, I worked on projects involving participatory sensing campaigns using environmental sensors in urban settings. I developed models to temporal heatmaps of the pollution in space and time. Those pollution heatmaps served as inputs for our Healthy route recommendations for city dwellers, which were calculated based on self developed multiple objective shortest path algorithms.

Career at AGT International

Basketball Analytics for the Final Four

I led sport analytics projects, including instrumenting professional basketball players to monitor their performance. I developed data-driven stories and analytics for publishing companies, enhancing audience engagement during sports events.

Career at AGT International

Basketball Analytics for the Final Four

I led sport analytics projects, including instrumenting professional basketball players to monitor their performance. I developed data-driven stories and analytics for publishing companies, enhancing audience engagement during sports events.

Career at AGT International

Professional Bull Riders (PBR)

I led the development of an AI-based scoring system for PBR. By implementing sensors on bulls, I built models to objectively score bull riding performances, addressing biases in the judging process and optimizing the model through continuous feedback from judges.

Career at AGT International

Professional Bull Riders (PBR)

I led the development of an AI-based scoring system for PBR. By implementing sensors on bulls, I built models to objectively score bull riding performances, addressing biases in the judging process and optimizing the model through continuous feedback from judges.

Proxify

Bef4U Project

Bef4U, a company specializing in energy efficiency for industry and large tertiary sectors, contacted me to help build a machine learning model to detect electric consumption anomalies. I designed an online approach based on a sliding training and testing window. In each step, an XGBoost model was trained, and a z-score was calculated for anomaly detection, improving energy efficiency monitoring for shopping malls.

Proxify

Bef4U Project

Bef4U, a company specializing in energy efficiency for industry and large tertiary sectors, contacted me to help build a machine learning model to detect electric consumption anomalies. I designed an online approach based on a sliding training and testing window. In each step, an XGBoost model was trained, and a z-score was calculated for anomaly detection, improving energy efficiency monitoring for shopping malls.

Proxify

Electrolux Project

Electrolux, a global leader in household appliances, aimed to forecast service call volumes during the warranty period. I developed a regression model using Markov Random Fields to predict the service call ratio (SCR) based on one year of historical data, achieving a 4.2% Mean Absolute Percentage Error (MAPE) for the food preservation category. The project successfully provided precise service call forecasts, enhancing customer service efficiency.

Proxify

Electrolux Project

Electrolux, a global leader in household appliances, aimed to forecast service call volumes during the warranty period. I developed a regression model using Markov Random Fields to predict the service call ratio (SCR) based on one year of historical data, achieving a 4.2% Mean Absolute Percentage Error (MAPE) for the food preservation category. The project successfully provided precise service call forecasts, enhancing customer service efficiency.

Proxify

Magenta Insight LTD

Magenta Insight LTD sought an AI solution to detect when water is pumped from wells using offline temperature sensors. I developed a Python-based system comprising a dashboard for visualization and parameter tuning, and a batch processing engine for analyzing sensor data. The AI model learned to identify pump usage by detecting temperature changes when water was pumped,

Proxify

Magenta Insight LTD

Magenta Insight LTD sought an AI solution to detect when water is pumped from wells using offline temperature sensors. I developed a Python-based system comprising a dashboard for visualization and parameter tuning, and a batch processing engine for analyzing sensor data. The AI model learned to identify pump usage by detecting temperature changes when water was pumped,

Proxifiy

Your Political Monitor

Your Political Monitor, led by CEO Caspar Pelle, aimed to create an advanced AI-powered web app for summarizing political messages. I was tasked with developing a demo web app where users could input text or upload PDFs to receive concise summaries without relying on Chat GPT. After researching and selecting a suitable pre-trained LLM model, I implemented LangChain's Map-Reduce and Refine techniques for handling larger documents. The app allows users to enter text, generate summaries, and view details on word count and processing time, providing accurate and timely summaries of political content.

Proxifiy

Your Political Monitor

Your Political Monitor, led by CEO Caspar Pelle, aimed to create an advanced AI-powered web app for summarizing political messages. I was tasked with developing a demo web app where users could input text or upload PDFs to receive concise summaries without relying on Chat GPT. After researching and selecting a suitable pre-trained LLM model, I implemented LangChain's Map-Reduce and Refine techniques for handling larger documents. The app allows users to enter text, generate summaries, and view details on word count and processing time, providing accurate and timely summaries of political content.

Testimonials

Testimonials

  • Duncan McNicholl

    Co-founder & CEO at

    Uptime Global

    In the discovery phase, I collaborate closely with stakeholders to understand their goals and challenges, while gathering and exploring relevant datasets to define the problem statement and key metrics for success.

  • Caspar Pille

    Co-founder & CEO at

    Polpo

    Through meticulous data cleaning and exploratory analysis, I refine the project scope and select appropriate modeling techniques, ensuring clarity on the problem definition and the methodology to address it effectively.

  • Christos Tsirogiannis

    co-worker at

    Polpo

    I am really glad I had the opportunity to work with Roel! He is an exceptional person in terms of communication and apparently hard work. Definitely a problem solver and very independent worker regarding the tasks assigned. Overall and most importantly a very kind and fun person, always open to new challenges. It was a great opportunity to meet and learn from him.

01

Discover

In the discovery phase, I collaborate closely with stakeholders to understand their goals and challenges, while gathering and exploring relevant datasets to define the problem statement and key metrics for success.

Develop

During development, I focus on training and refining models iteratively, optimizing their performance and usability while concurrently designing intuitive user interfaces to facilitate understanding and utilization of the insights generated.

03

Define

Through meticulous data cleaning and exploratory analysis, I refine the project scope and select appropriate modeling techniques, ensuring clarity on the problem definition and the methodology to address it effectively.

02

Deliver

In the final phase, I ensure the successful deployment of the solution, monitoring its performance over time and providing comprehensive documentation and reporting to stakeholders to support decision-making and ongoing maintenance efforts.

04

Specialization

Specialization

Predictive Maintenance

Sports Analytics Based on Accelerometer and Gyroscope Sensors

Image Classification Using RGB and Hyperspectral Imaging

Machine Learning with Regression Techniques for Prediction (e.g., Survival Probability, XGBoost)

Anomaly detection

Multiple aspect shortest paths

Predictive Maintenance

Sports Analytics Based on Accelerometer and Gyroscope Sensors

Image Classification Using RGB and Hyperspectral Imaging

Machine Learning with Regression Techniques for Prediction (e.g., Survival Probability, XGBoost)

Anomaly detection

Multiple aspect shortest paths

Predictive Maintenance

Sports Analytics Based on Accelerometer and Gyroscope Sensors

Image Classification Using RGB and Hyperspectral Imaging

Machine Learning with Regression Techniques for Prediction (e.g., Survival Probability, XGBoost)

Anomaly detection

Multiple aspect shortest paths

My Values

My Values

Innovative. We constantly seek out new and better ways to solve problems.

Innovative. We constantly seek out new and better ways to solve problems.

Innovative. We constantly seek out new and better ways to solve problems.

Collaboration. I work to achieve your goals and celebrate our successes.

Collaboration. I work to achieve your goals and celebrate our successes.

Collaboration. I work to achieve your goals and celebrate our successes.

Empathy. I put myself in the shoes of our customers to understand their needs.

Empathy. I put myself in the shoes of our customers to understand their needs.

Empathy. I put myself in the shoes of our customers to understand their needs.

Reliability. As the cornerstone of trust, ensuring consistency, accuracy, and dependability in all endeavors.

Reliability. As the cornerstone of trust, ensuring consistency, accuracy, and dependability in all endeavors.

Reliability. As the cornerstone of trust, ensuring consistency, accuracy, and dependability in all endeavors.

Customer obsession. I am passionate about helping my customers succeed.

Customer obsession. I am passionate about helping my customers succeed.

Customer obsession. I am passionate about helping my customers succeed.

Transparency. I believe in communicating openly and honestly with everyone.

Transparency. I believe in communicating openly and honestly with everyone.

Transparency. I believe in communicating openly and honestly with everyone.

Frequently asked questions

Frequently asked questions

Predictive Maintenance

Sports Analytics Based on Accelerometer and Gyroscope Sensors

Image Classification Using RGB and Hyperspectral Imaging

Machine Learning with Regression Techniques for Prediction (e.g., Survival Probability, XGBoost)

Anomaly detection

Multiple aspect shortest paths

Predictive Maintenance

Sports Analytics Based on Accelerometer and Gyroscope Sensors

Image Classification Using RGB and Hyperspectral Imaging

Machine Learning with Regression Techniques for Prediction (e.g., Survival Probability, XGBoost)

Anomaly detection

Multiple aspect shortest paths

Predictive Maintenance

Sports Analytics Based on Accelerometer and Gyroscope Sensors

Image Classification Using RGB and Hyperspectral Imaging

Machine Learning with Regression Techniques for Prediction (e.g., Survival Probability, XGBoost)

Anomaly detection

Multiple aspect shortest paths

Besides programming

Besides programming

Outside of programming, I find immense joy in spending quality time with my family, as they are my greatest source of happiness and inspiration. In addition to this, I am deeply passionate about teaching Hatha yoga and brewing homemade kombucha. Embracing these hobbies allows me to cultivate mindfulness, wellness, and creativity in my daily life.

Outside of programming, I find immense joy in spending quality time with my family, as they are my greatest source of happiness and inspiration. In addition to this, I am deeply passionate about teaching Hatha yoga and brewing homemade kombucha. Embracing these hobbies allows me to cultivate mindfulness, wellness, and creativity in my daily life.

Outside of programming, I find immense joy in spending quality time with my family, as they are my greatest source of happiness and inspiration. In addition to this, I am deeply passionate about teaching Hatha yoga and brewing homemade kombucha. Embracing these hobbies allows me to cultivate mindfulness, wellness, and creativity in my daily life.

“Give the ones you love, wings to fly, roots to come back and reasons to stay”

“Give the ones you love, wings to fly, roots to come back and reasons to stay”

Dalai Lama XIV

Dalai Lama XIV

Roel Heremans
senior data scientist

Schedule a call with me

Roel Heremans © 2024.

Roel Heremans
senior data scientist

Schedule a call with me

Roel Heremans © 2024.