Description

Mathematical engineer (MSc, magna cum laude) with 6 years designing and deploying end-to-end AI systems from vision to large language models: AI-powered waste-stream analytics, multimodal similarity search, on-edge inference, and LLM-driven document intelligence, and 8 years full-stack development. Expert with Python, PyTorch, LLMs, Langchain, OpenCV, YOLOv9/v7, Pandas, Numpy deploying on AWS SageMaker, Azure ML and GCP.

Languages

Dutch
Native or bilingual
English
Native or bilingual
French
Basic

Workplace preferences

Can work on-site

Brussels (up to 50km)

We Make Magic
ML Engineer
DIGITAL AND IT
April 2020 - Today (6 years and 2 months)
Hasselt, Belgium
• Rematics: waste‑stream object detection & tracking (AWS SageMaker, Python)
Architecture design and deployment of an end‑to‑end training -> edge‑inference
pipeline for detecting and tracking waste materials on multiple garbage‑truck
camera rigs; supports rolling model re‑training, CI/CD‑based promotion, and
on‑vehicle OTA updates (YOLOv9, ByteTrack, AWS SageMaker Pipelines, Lambda, Angular monitoring dashboard)

• Decomecc: Architecture and implementation of a pipeline to detect coil data from invoices using LLMs (Azure, LLMs, pattern matching, prompt engineering)

• Veolia & Vanheede: classification + tracking of mixed waste
Fine‑tuned YOLOv7 models with extensive data‑augmentation and class‑imbalance sampling; integrated DeepSORT for multi‑frame ID continuity.

• Entiris: matching samples to original wallpapers using location matching (OpenCV, SIFT, FLANN) and deep learning techniques (Contrastive learning, Resnet architectures)

• Biostoom: recyclable‑fraction classifier
Adapted the Veolia/Vanheede pipeline to Biostoom’s setup; achieved 95 %
classification accuracy on hold‑out test set of industry‑specific waste fractions.

• OVAM: Help in the creation of a demo for automatic matching of waste streams with processors (LLMs, agentic AI, LangGraph, Pydantic)

• S-Lim: Implementation of an AI chatbot, smart AI search and user reporting of illegal dumping with photos using LLMs for classification in the S-Lim app (LLMs, RAG, Azure AI Search, Azure Bot Service)

• Energie.be: Help in the creation of a custom chatbot solution (LLMs, Langchain, RAG, Chromadb, prompt engineering, red teaming)

• Amfori: Sensitivity analysis and fraud detection of the reports in the sustainability platform (NLP, text embeddings, sentiment analysis, statistical analysis and data visualization)

• Amfori: Detection of high risk monitorings in the quality monitoring project (outlier detection on tabular data using SVMs, DBScan and isolation forests with a Kubeflow pipeline in Vertex AI)
Python AWS Pytorch Generative AI Computer Vision
We Make Magic,
MOBILE IOS & ANDROID DEVELOPER
DIGITAL AND IT
November 2016 - Today (9 years and 7 months)
Leuven, Belgium
Design and development in iOS (Swift, Objective C), Android (Kotlin, Java), web
(Javascript, React, Python, Flask), databases (MySQL, PostgreSQL, Redis, Cloud
Firestore):

• Voka Networking app shows you all the latest Voka events in a calendar and
helps you manage your network.

• Medbook is an app for medical students and residents to evaluate procedures
and keep track of their learning process

• Collabor8 is a tool developed for experiental marketing activations with
calendars, events, notifications, location tracking, material tracking, messaging,
contract signing, profile page, etc.

• Ekho is used by tens of thousands of people in Belgium to compare opinions
during live football matches on RTBF and The Voice Belgique.

• BioRICS - Stress at Work uses Bluetooth-connected smartwatches for heart rate monitoring to estimate stress levels
Android iOS Swift Kotlin Java
HabitBull
Founder
DIGITAL AND IT
April 2014 - June 2016 (2 years and 2 months)
London, United Kingdom
Technical and business development of an Android and iOS app for habit tracking.

• 2M+ downloads (Android), 500k downloads (iOS), 140k active users/month
• ‘Editor’s Choice’ on Google Play / featured in “Get Things Done” category
• 4.5/5 average rating over 50k+ reviews
• Featuring near-real-time data sync and in-app forums
• Technologies: native iOS (Swift) and Android (Java), backend in AWS with MySQL and MongoDB with a node.js web api
• Revenue generating / acquired
Android iOS Business development Marketing Swift