Technische Universität Berlin + Ghent University (Double PhD)

Technische Universität Berlin

Ghent University

Klein Seminarie Roeselare

Image Processing and Machine Learning for VR and Medical Applications, PhD

Datameister

Founder

Independent consultant in the areas of Image Processing,  AI and NLP.

DM-OG

Independent Consultant

Oqton, Inc

Senior AI Engineer

Enable marketing, sales and support teams to make better hyper-personal customer processes, by adding contextual and behavioural intelligence to the apps they use.

journy.io

Chief Data Scientist

We believe artificial intelligence and other technological advancements are driving medical science breakthroughs and will reshape the way cancer is diagnosed and treated. As a bio-informatics company, Hyperfusion.ai stands at the vanguard of applying AI to DCE-imaging, with a mission to give patients better and faster feedback on the treatment he/she is receiving.

Hyperfusion.ai

Co-Founder and Medical AI Vision Expert

- Became co-director of the underlying non-profit organization TEDDicts VZW.
 - Learned how to manage and motivate a large group of 30 volunteers, through investing into a lot of personal informal contacts and organizing team-building moments.
 - Gained experience in dealing with various constraints, especially budget-wise
 - Followed up and tried to optimize the collaboration between all teams: communication, partnerships, event, and content team.
 - Learned how to set-up a strategy for involving our audience, i.e. the TEDx community. Resulted in organizing a number of smaller low threshold and free (as in beer) events. Not mentioned below.

 - Together with the event manager successfully setup a very meaningful, cost-efficient and experimental cooperation between content-creating companies (Nøcomputer, klankwerk, MPATHY) and AV providers (Barco, IOSONO), to allow for a very experimental AV setup with generative designs, sound-scape music and semi-immersive experience. As such, we helped Barco to experiment on how to deliver events to audiences not present at the venue. (TEDxGhent main event #6) 

 - Practical coordination of the following events:
  * TEDxGhentLive: TEDxCERN (live-stream, +- 200 attendees)
  * TEDxGhent - UGent: PhD Contest (10 contestants, +- 100 attendees)
  * TEDxGhentSalon: Women & STEM (4 speakers, +- 150 attendees)
  * TEDxGhent #6: Bits of Love (Main Event, 600 attendees)

TEDxGhent

Co-Curator

- Leading and motivating the Event Team of 8 volunteers.
 - Coordination of the technical production (AV captation), venue matters (catering, stage design, crowd flow) and side-entertainment (workshops, interactive demos, Jenga-playing robots, ...). For the following list of smaller and one larger 
  * TEDxGhentLive: TEDGlobal - South (live-stream, +- 150)
  * TEDxGhentSalon: Women in STEM (4 speakers, +- 250)
  * TEDxGhentSalon: Reasonable Doubt (4 speakers, +- 200 attendees)
  * TEDxGhent - UGent: PhD Contest (10 contestants, +- 100 attendees)
  * TEDxGhent #5: Celebration (Main Event, 18 speakers, +- 800 attendees)

Event Manager

- Research topic: exploring radical new ways to code imagery, based on the recent advances in machine learning. Techniques explored included kernel-adaptive filters, RBF networks and Mixture-of-Experts in various forms. This led to the development of the Steered Mixture-of-Experts (SMoE) framework for universal coding. As such, through the analytical continuous form of our model, the approach is able to deliver extra functionality at code level (e.g. super-resolution, denoising, feature-descriptors, ...), while being bit-efficient. The framework scales especially well to higher dimensional image data, such as light fields.

This resulted in invited talks at Google (Zürich and Mountain View), Disney and Netflix. The Google Faculty Award 2016 was awarded to TU Berlin for the continuation of this work.

 - Doctoral Schools (Specialist Courses): Machine Learning, Bayesian Statistics, Computational Intelligence
 - Doctoral Schools (Soft Skills Courses): Communication Skills, Conflict Management, and the workshop "From PhD to labor market'"
 - Obtained experience in writing project proposals (IWT, FWO).
 - Supervised lab sessions for the courses Multimedia and Multimedia Technologies.
 - Supervised master theses in the domains of machine learning, image coding, and DNA compression.

Double PhD Degree- Computer Science

- In charge the side-entertainment: coming up with ideas, coordination, and communication. This involved delivering sandwiches with drones, spoken work poetry, workshops, ... 

 - Part of the content team coming up with suggestions for speakers, and following up on the communication and speaker coaching. (TEDxGhent #4 Main Event, +- 400 attendees)

Spices Manager

(Same as for Ghent University)

My PhD is a very tight collaboration between Ghent University - iMinds and TU Berlin in the form of a Double PhD. 

 - Research topic "Learn to compress": exploring radical new ways to code imagery, based on the recent advances in machine learning. Techniques explored included kernel-adaptive filters, RBF networks and Mixture-of-Experts in various forms. This lead to the development of the Steered Mixture-of-Experts (SMoE) framework for universal coding. As such, through the analytical continuous form of our model, the approach is able to deliver extra functionality at code level (e.g. super-resolution, denoising, feature-descriptors, ...), while being bit-efficient. The framework scales especially well to higher dimensional image data, such as light fields.

Double PhD Degree - Computer Science

Working towards a publication and doing further research on my master's thesis: "Image Coding using Adaptive Sampling and Data-Adaptive Kernel Regression".

Researcher

- 2012: Research and successful development of C++ and CUDA implementations for automatic image camera noise assessment in photos for MostPhotos.com.

 - 2011: Development of Google Analytics integration and database design as a subcontractor for Teamleader.be (then: rgbscape.be) (PHP, MySQL).

 - 2010-2011: In-house development of a web CRM system integrating with existing custom Visual Basic software for an industrial cleaning firm De Mandel BVBA (PHP, MySQL).

 - 2008-2009: Working with small and medium sized companies across Belgium and the Netherlands providing total or partial development of company websites and back-ends. Typically involving parsing and collecting RSS feeds (PHP, MySQL).

Ninetynine

Self-employed

Co-founded a t-shirt label including setting up logistics, custom webshop, photo shoots, printing, and local retail cooperations.

Ruben Verhack