Work experience

• Give a Day March 2020 to present
  Give a Day is a scale-up in Belgium that built a CRM for volunteer organisations and that hosts the largest volunteer matching platform in Belgium. Recently Give a Day launched the 'neighbor phone', connecting neighbors to help each other with small tasks or to perform a social activity.
  Position Machine learning engineer and full stack developer
  Achievements We developed custom algorithms for the automatic recognition of voicemails, predicting the optimal time to call a user, and to perform optimal matching between neighbors. We developed a responsive SPA that is a full-fledged CRM for volunteers, integrating management, planning, communication, expenses and contracts.
  Languages and frameworks

  • Machine learning: python + numpy + sklearn + scipy + librosa
  • Frontend: nuxt + vue
  • Backend: nodejs + MariaDB + google TTS
• BCG - GAMMA March 2020
  BCG - Gamma is a global data science consulting company.
  Position Full stack developer
  Achievements We developed an easy-to-use application for call centre employees that ranks prospects automatically based on previous interactions. The application improved the customer acquisition rate with a predictive model that could single out promising prospects.
  Languages and frameworks

  • Frontend: Vue
  • Backend: Django + MYSQL
  • Machine learning: sklearn
• Stepwise November 2018 to February 2020
  Stepwise is a start-up in Belgium that develops a tailored training program for novice and experienced runners.
  Position Founder, machine learning engineer and full stack developer
  Achievements I developed two AI models, one that creates a personalized training program for a runner based on their personal characteristics, and one low-latency, on-device model that would estimate the actual speed and position of the runner, even when faced with spotty GPS and sudden speed changes.
  Languages and frameworks

  • Machine learning models: Rust + Java + Logistic regression + Dynamic programming
  • Mobile apps: Vue + Java + Objective C + Rust
  • Backend: Java + MYSQL
• COLDSTART October 2019
  ColdStart is an AI start-up that helps people to manage their online personal data.
  Position Frontend developer
  Achievements In only two weeks we developed the first MVP for ColdStart, allowing them to rapidly gather feedback from their test users.
  Languages and frameworks

  • Frontend: Vue
  • Backend: Django + MYSQL
• COLRUYT July 2018 to October 2018
  Colruyt is a supermarket chain in Belgium with a focus on healthy and environmental friendly foodstuffs.
  Position Functional analyst and full stack developer
  Achievements We developed an application that allowed Colruyt personnel to manage the complex footprint data associated with food production, reusing and augmenting existing data. The solution accurately computes an “eco-score” for a large number of foodstuffs, taking into account the carbon footprint of the entire chain, from farmers to supermarkets.
  Languages and frameworks

  • Frontend: Angular
  • Backend: Java + Oracle SQL
• CONSORTIUM OF SOCIAL SECRETARIATS March 2018 to June 2018
  Position Functional analyst and project proposal writer
  Achievements In Belgium various functions of payroll are handled by ‘social secretariats’. We developed a multi-year project proposal for a consortium of social secretariats to develop a shared back office that could be used by all consortium members.The solution employed a low-code platform to substantially reduce the cost and improved the speed of delivery of the solution.
• MICROSOFT December 2017 to February 2018
  Position Blockchain developer
  Achievements We developed an easy-to-use wizard in Xamarin (Android, IOS, UWP) to fully specify a blockchain application, and automatically generate a solidity contract or enterprise smart contract and the necessary configuration files to deploy these contracts to the new Azure cloud solution, App Builder.
  Languages and frameworks

  • Develop wizard UI: C# + Xamarin Forms
  • Blockchain: solidity + ethereum
• KPMG NETHERLANDS June 2017 to December 2017
  KPMG is a professional service company and one of the Big Four auditors.
  Position Machine learning engineer and blockchain developer
  Achievements We developed an application that combines an artificial intelligence algorithm with a blockchain to drastically reduce the manual work needed to reconcile financial transactions from multiple sources when auditing SPV's.
  Languages and frameworks

  • Store transactions on blockchain: Corda
  • Machine learning: Java + clustering algorithms + string distance metrics
  • Web: Angular 4 + Java
• AMSTERDAM TRADE BANK March 2017 to May 2017
  Amsterdam Trade Bank is a Dutch bank that focuses on trade financing for companies wishing to do business in Russia and former Soviet states.
  Position Blockchain developer
  Achievements We developed a blockchain application that allowed for the tracking of goods from producer to consumer, including the financing of the goods.
  Languages and frameworks

  • Blockchain: Ethereum
  • Frontend: Angular 2 application
  • Backend: Java + web3j
• KBC January 2017 to March 2017
  KBC is one of the largest bank and insurance companies in Belgium and active in Central and Eastern Europe.
  Position Web developer
  Achievements We developed an application that allowed for the easy tracking of a car sale and the car's insurance by the customer, the car salesperson and KCB insurance, radically speeding up the process and reducing the amount of paperwork.
  Languages and frameworks

  • Blockchain: Hyperledger
  • Frontend: Angular 2
  • Backend: Nodejs + express
• SCRIPTBOOK December 2015 to November 2016
  ScriptBook is a start-up that develops text analysis algorithms to forecast the commercial and critical success of movies based on a multi-level analysis of the screenplay.
  Position CTO and machine learning engineer
  Achievements In 1 year we developed a prediction system that outperforms human experts and other automatic systems for predicting the US box office of a movie, based on the screenplay of the movie and the release strategy for a movie. The product is currently being tested by a major US studio, a US film funds and a European distributor.
  Languages and frameworks

  • Machine learning: python + auto-scaling AWS instances + numpy + sklearn + nltk + lda
  • Website: HTML 5 + angular JS + flask
• NOISYCHANNELS July 2015 to November 2015
  NoisyChannels is a start-up that develops text analysis algorithms to assist knowledge workers in operational processes.
  Position Machine learning engineer and full stack developer
  Achievements in 3 months we developed two separate solutions for two of the biggest social secretariats in Belgium and one large insurance company, resulting in the first sale for NoisyChannels. We developed custom classification algorithms of customer emails based on the contents of the e-mail and attachments, and a crawler and automatic classifier of news messages for social secretariats.
  Languages and frameworks

  • Machine learning: java + Stanford CoreNLP + tesseract OCR + apache PDFBox
  • Web: HTML 5 + angular JS + twitter bootstrap
• CRELAN ONLINE April 2015 to July 2015
  Crelan is a retail bank and insurance company with over 1.700 employees and 430.000 customers in Belgium.
  Position Web developer
  Achievements I joined team halfway in a project to develop the MIFID questionnaire in Crelan Online and Agent Portal. The project was running late, but we achieved the deadline and managed to add several unbudgeted features.
  Languages and frameworks

  • Backend: java + struts + Oracle RDBMS
  • Frontend: HTML5 + jquery
• TERASHARE.NET April 2014 to August 2014
  Terashare is a P2P application that allows people to send large files for free. It uses the bittorrent protocol to send files reliably, fast and secure.
  Position Full-stack developer
  Achievements We successfully developed a user-friendly application (Windows, MacOS & Linux) that uses the bittorrent protocol to send files (up to several terabites) for free. The application also performs UDP-relaying and NAT hole punching, to guarantee connectivity with a P2P protocol.
  Languages and frameworks

  • P2P application: python + pytorrent + nsis
  • Website: angular + python + django + mysql
• VOKA December 2013 to April 2014
  VOKA (Vlaams netwerk van ondernemingen) is an employers' organization in Flanders, representing over 18.000 companies in Flanders and Brussels.
  Position Backend developer
  Achievements We radically improved the speed of the existing voka.be website, with more than 10x average speed-up. The loading time of the slowest pages went from 3 minutes to 2 seconds.
  Languages and frameworks

  • Umbraco CMS + many custom .NET modules
• RESTO.BE Juli 2013 to November 2013
  Resto.be is the largest restaurant site in Belgium, with more than 3000 affiliated restaurants.
  Position Full-stack developer, scrum master
  Achievements We expanded the existing website to improve the search results for user by taking into account their current location.
  Languages and frameworks

  • Location based search: apache solr
  • Web: java + spring + MS SQL
• DE TIJD December 2012 to June 2013
  ‘De Tijd’ is a Belgian newspaper that focuses on business and economics.
  Position Full-stack developer, technical lead
  Achievements We modified the editorial website used by the journalists and the REST API and rending engine for the new tablet app. The app was then considered by many the best tabled app for any newspaper in Belgium.
  Languages and frameworks

  • Website: java + spring + thymeleaf + oracle SQL
  • REST API: java + spring
• RESTO.BE April 2012 to November 2012
  Resto.be is the largest restaurant site in Belgium, with more than 3000 affiliated restaurants.
  Position Front-end developer
  Achievements We developed a new javascript application to streamline reservations for restaurants and to perform table management.
  Languages and frameworks

  • Web: java + spring + MS SQL + dojo + javascript canvas
• CULT August 2011 to April 2012
  Cult was a startup based in San Francisco that developed a location based social network, connecting people in a certain location (e.g. the Guggenheim museum) to other people in similar locations around the world.
  Position Machine learning engineer, full-stack and mobile developer
  Achievements I developed large scale clustering algorithms, scraping algorithms, a backend and frontend and mobile app for the new social network app.
  Languages and frameworks

  • Large scale clustering: java + cassandra
  • Web: HTML5 + javascript + python + django
  • iOS app: objective c
• Katholieke Universiteit Leuven Sep 2005 to April May 2011
  KULeuven is the largest university from the Benelux and was the first university in Europe with a degree of civil engineer in computer science.
  Position Researcher, PhD student end post-doctoral researcher
  Achievements My research field was Natural Language Processing and I focused on two different topics: (1) combining text analysis with analysis of images and video's, and (2) improving existing NLP methods with weakly-supervised training. I developed a new language model (the Latent Words Language Model) that beat all state-of-the-art models at the time.
• Verbal Vision Feb 2004 to present
  Verbal Vision is an NGO that helps organisations in their communication efforts towards a diverse audience, with a focus on minority and immigrant groups. Since 2015 Verbal Vision is focusing mostly on publishing ultra-diverse children books under the label Studio Sesam.
  Position Member of the board