Pieter van Beek

In the context of a European public request for offer (RFO), I wrote the technical specifications document. This required designing multiple 4-petabyte and 50,000 SpecInt rate high-performance compute and storage clusters with various commodity hardware setups.

Invented and implemented a provably and verifiably secure method for exchanging data within dynamic user groups wherein all users can invite new users. The method is to be used in group chat applications and electronic medical records, and also in military, police, and intelligence communities.

Designed and helped implement a multi-tenant Hadoop hardware and software infrastructure for Vancis, an internet service provider. The infrastructure used Xen and VMWare for hardware virtualization, Hortonworks, and EMC’s Pivotal solution for long-term storage of cool data.

A peta-scale data storage and exchange for academic researchers.

Confronted with the task of implementing a RESTful web service in Ruby, I checked out the existing frameworks, but then decided to create my own since I couldn't find a framework with all of the following properties:

SMALL. Some of these frameworks are really big. If understanding the framework takes more time than writing my own, then that framework had better be more powerful than what I can come up with by myself.

NO EXTENSIVE TOOLING OR CODE GENERATION. With Ruby, you shouldn't need code generation.

FULL SUPPORT FOR HTTP/1.1 CONDITIONAL REQUESTS. Most libraries’ support is limited to "if-none-match" and "if-modified-since" headers, and only for GET and HEAD requests. For RESTful web services, the "if-match" and "if-unmodified-since" headers are at least as important, particularly for unsafe methods like PUT, POST, PATCH, and DELETE.

RESOURCE-CENTERED. Some libraries claim RESTfulness, but at the same time have a servlet-like interface, which lets you implement method handles like doPOST(URL), in which you have to find out what the resource is posted to, depending on the URL. This library lets you focus on your server–side resources and their representations.

Built an online, Flash-based game for the University of Amsterdam and a television broadcasting company that allowed academic economists to observe and experiment with human behavior in cooperative games.

A web portal for efficiently running common large-scale computations (e.g., BLAST) on the LHC Compute Grid with tens of thousands of compute nodes and petabytes of storage.

Co-developed a web portal for performing large-scale inverse modeling, machine learning, and parameter optimization on LSG and LISA.

Ported a computationally intensive tomographic reconstruction algorithm from CPUs to GPUs (Nvidia Tesla cards) with Nvidia's CUDA and cuFFT toolkits. This was while working at SARA, the Dutch National Supercomputer facility, in collaboration with the University of Utrecht, Netherlands.

Co-developed a literature–based keyword enrichment tool for microarray data analysis.