The Magic of Machine Learning


Any sufficiently advanced technology is indistinguishable from magic

Arthur C. Clarke

Making the Magic Happen


I am passionate about technology and innovation. In 2017, looking for a new intellectual challenge, I started on a journey into the magical realm of Machine Learning.

Since then, I have learned that Machine Learning is not always the right or only solution. It always starts with understanding the business problem and context. Luckily this is one of the main skills I picked up earlier in my career.

Sometimes a Machine Learning model can be used. Most of the times business rules are enough or can improve the output of the models. I'd like to help you designing and building the right solution for your particular problem, not the most hyped one.

Next to that, you also need to make sure the solution is adopted by your team. For this proper attention to change management is needed. Managing resistance to change. Explaining how the new technology works and how it should be used. These skills are also part of my toolkit, gathered through managing a broad variety of project and business teams.

Some projects


  • Prediction engine for localization of assets using Bluetooth tags, including pre-processing of the data, prediction based on different algorithms and post-processing of the results. Creation of powerful business insights based on the generated data. Using different ML and AI technologies.
  • Proof of Concept for prediction of missing classification fields in article master data, using a text matching algorithm.
  • Development and deployment of a web application that connects with Bluetooth sensors, and processes and presents the data.
  • Development and deployment of a web service for a rule based algorithm that will suggest the optimal surfboard based on user preferences.
  • iOS App and underlying Software Package for skin blurring and object recognition of patients and medical staff in videos, using Computer Vision for skin blurring and Deep Neural Networks for face, pose and object recognition.
  • Document Classification of insurance documents and emails, using Computer Vision and Deep Neural Networks.
  • Proof of Concept for Digit Recognition for service vouchers, using Computer Vision and Deep Neural Networks.
  • Proof of concept for Document Separation and Classification for building permits, using Deep Neural Networks and Natural Language Processing.
  • Proof of Concept for Sales Forecasting using Neural Networks and Natural Language Processing.
  • Contact Me


    Please find my contact details below for an initial chat. Looking forward to making the magic happen.

    Stefaan Debevere

      

    Awtentik BV - Lijsterdreef 12 - 8490 Varsenare - Belgium - BE0546.660.623