Sikando finally gives me the chance to look for interesting projects without being limited by the goals and vision of an employer. I develop prototype software, one-of-a-kind embedded real-time software, develoment tools and other challenging kinds of software. More details on the Sikando website http://www.sikando.com.In my experience, there are some very creative people around with have great ideas for projects that can provide a significant added value, but lack the time or know-how to figure out how these can be implemented in software. I really love to work on such projects, using all my experience and common sense to find a practical solutions that works, and evaluate alternatives where appropriate. If you have such an idea, please do not hesitate to contact me to arrange a meeting to explore the possibilities.

Embeetle is an entirely new C/C++ IDE focused on microcontroller coding only. It’s not yet another plugin for Eclipse, Atom, Visual Studio or any other existing IDE. We’ve started from scratch, and developed only what the embedded coder needs. Clean and simple. Refreshing.https://embeetle.cominfo@embeetle.com

FMTC is a research organisation for mechatronics (combination of mechanics and electronics, related to factory automation) for Flemish SME's. When I started working here, both FMTC and I assumed that this involves a lot of embedded and other software development, so this looked like a perfect opportunity to broaden my software know-how.To some extent, this was shown to be true. At FMTC, I have worked on very diverse projects, ranging from Eclipse plug-in development in Java over the integration of networking code in C with Matlab to real-time control code with Xenomai, Orocos and EtherCAT I/O and interfacing over serial line to ZigBee low power radio. I have even done some FPGA development and experimented with contactless power transfer over a resonant magnetic circuit. As the last two topics indicate, there was not always enough software implementation work to keep me busy.My most prominent project at FMTC was certainly the software integration for the world's one and only Badminton robot: see http://www.youtube.com/watch?v=LSax71cn6A4. B.t.w, the person in this video is a colleague, I am probably sitting behind the computer.

Time to take some time off, after leading a busy professional and family life for over 20 years.Also time to work on the idea of using a data flow representation as internal data structure for compiler optimization. This proved much more challenging then initially anticipated, especially because my aim was to accept arbitrary C code (including pointers, dynamic allocation, goto and longjump ...) or C++ code at the input. As a front end, I tried EDG, gcc and llvm. Llvm wins.During the same period, I experimented with the idea of starting my own company (Sikando). This was a very valuable experience, and a big step from my previous research and development work. However the combination of this big change with the challenging work on the software side was too much too handle at this time, so I put this idea on hold and accepted a new job.

After working in a commercial company for nearly 10 years, I felt the need to do some more research oriented work. At that time, IMEC was still actively working on chip design methodology and tools, which was (and is) still my area of interest.One project that I am proud of is the development of TIPSY, a C++ library to create concurrent timed executable models of digital hardware, comparable to but predating behavioral SystemC. A second version of TIPSY was also able to model software components, including real time operating systems (see http://downloads.hindawi.com/journals/asp/2007/075373.pdf, http://publik.tuwien.ac.at/files/pub-et_11130.pdf and http://www.patents.com/us-6952825.html). This functionality has been offered to the SystemC standardization committee, but was never accepted. On an unrelated note, remark that SystemC never realized its promise of becoming the single language of choice for system and hardware design ;-)My most interesting project at IMEC was the development of SPRINT, a tool to parallelize sequential C code based on user directives (see http://www.hindawi.com/GetArticle.aspx?doi=10.1155/2007/75373&e=cta). This gave me the opportunity to explore high level compiler optimizations and automatic code transformations for parallelisation. Sadly, IMEC did not give me time to replace its internal control flow graph data structure with a much more flexible and powerful data flow representation. Eventually, IMEC stopped all research on design methodologies and tools, so I decided to leave.

When I started working at EDC, the company was actually called Silvar-Lisco, and had been market leader in Standard Cell and Gate Array design tools for many years. Working at Silvar-Lisco did however give me the opportunity to improve my programming skills, in terms of programming languages (FORTRAN, C, C++) and software engineering (modularity, object orientation, version and configuration management for large scale software projects).After a few years, the European branch (Lisco) was renamed to EDC and sold to SCS, which was in turn bought by Mentor Graphics. EDC's main project was DSP Station, a set of High Level Synthesis tools for chip design. Input was a high level data flow language (Silage/DFL) and output was synthesizable VHDL or Verilog. The algorithms were based on research of the KULeuven and Philips Natlab (Philips Research Labs).Initial implementations of the tools were very unstable, and every attempt to remove bugs seemed to introduce others. My contribution, together with Erik Lesire, has been the design of a new core data structure (called SFG+), with well defined consistency rules that were enforced by pre- and postconditions and class invariants. These are well known concepts nowadays, but were new to the EDC engineering team at that time. SFG+ proved to be an enormous help to the stabilization of the initial prototype software. EDC has changed name and owner multiple times since then (Frontier Design, Adelante, ARM Belgium, Cognovo, U-blox), but still exists. SFG+ is still the core data structure of the tools that are now mostly used internally at U-blox. This work brought me in contact with high level compiler optimization techniques, a logical extension of symbolic manipulation. It also got me interested in data flow, which is in my opinion a much better foundation for high level optimization than the control flow graphs commonly used in compilers today.

ESAT is the Electrical Engineering department of KULeuven university. Although I studied electrical engineering, my interest in computer science was at least as strong. At that time (1983-1986), there was very or no knowledge of computer science at ESAT. Together with Luc Rijnders, I tried to introduce the concept of modular programming. It proved very difficult to get the idea across to my colleagues.So instead, I implemented a modular electrical simulator that could easily be reused by several colleagues in their research on various EDA tools, including waveform relaxation and high level timing verification. This proved very succesful. SIMMY, as the module was called, has been used at the university and at the IMEC research institute for more than ten years, long after I left. I am still proud of this result.Around the same time, I got interested in symbolic manipulation and optimalization of expressions, in casu the electrical equations of MOS transistors. Due to military service, I was not able to achieve any results then, but this topic has continued to draw my attention in different forms until today.