I'm useful everywhere where new things need to be understood, then
engineered into working prototypes and/or documented clearly.
I'm good at the creative process of moving from a vaguely formulated
problem to a working technical solution, and at the process of providing
clarity into complex algorithms and/or source code.
TECHNICAL EXPERIENCE DETAILS:
- Programming Languages
- ANSI C, C++, Java, Pascal, Perl,
assembler, FORTRAN (-77 and -90), Lisp, Smalltalk,
BASIC, VBA, UNIX shell scripts (sh/bash/ksh/C), Javascript, ...
- Operating systems and Platforms
- UNIX (Linux, HP-UX, AIX), Windows, Renesas M32192 with OSEK/VDX, OS/2,
DOS, VAX-VMS, Apple Macintosh, ...
- Software tools
- gcc, gdb, ddd, gprof, make, Microsoft Visual C++, Lauterbach debugger,
Vector tools, yacc/bison, (f)lex, vi(m), hexdump, UNIX tools
(grep/awk/sed/...), SVN/CVS/MKS/Clearcase, doxygen, javadoc, ...
- Libraries
- C++ STL, MFC, OpenGL, X-Windows, Java Sun container classes,
Java Swing, ...
- Communication protocols
- TCP/IP, HTTP, (E)SMTP, SNMP, CAN, SPI, ...
- Application languages etc.
- XML, XSL, SQL, mySQL, Postscript, Prolog, UML,
formal specification languages (Z, predicate calculus), ...
- Other relevant skills/knowledge:
- Good all-round knowledge in physics, electronics, mathematics
(incl. numerical and graph algorithms), AI fields (neural
networks, natural language processing, text association),
and computer science (hashing, random generation, balanced
trees, compiler design);
basic knowledge in information theory
- Computer science experience includes: Code generation (C/C++,
VBA, XSL, XML/HTML, Postscript); Coding of interpreters;
Creation of own software tools (e.g. Java file-scope class
detector, calling tree analyzers, BASIC variable definition
checker); Computer graphics (ray tracing, splines);
Object-oriented design (in any programming language); UNIX
system programming (pthreads, pipes, ...)
|
- 1992-1996: First jobs, Netherlands
- 1992 (Delft Institute of Micro-Electronics and Sub-micron
techology, Delft)
-- Writing a program for computation of charge distributions in
SiGe MOS transistors (C++, Apple Macintosh); plus assistance with
literature research.
- 1993-1994 (Warma Engineering, Ridderkerk bei Rotterdam)
-- General maintenance of a program (Basic, DOS) for
control/monitoring of heating installations.
- 1995 (INCORE Automatisering, Amsterdam)
-- Design and implementation of a program for generic and configurable
data communication (file transfer through TCP/IP) between a
warehouse management program and a remote database (C, OS/2).
- 1996 (TNO-FEL, Den Haag)
-- Implementation of a program for computation of the infrared
contrast of a navy ship, on the basis of a simple model. Also
assistance with making the model mathematically consistent.
(FORTRAN-90, DOS).
- 1996 (CMG, Den Haag)
-- Programming work for control of the mobile flood-protection
dam ("Stormvloedkering") near Rotterdam (UNIX, C/C++).
|
-
1996-1997 (CONSUL Risk Management B.V., Delft, with TU Delft,
Netherlands)
Research and software development for anomaly detection for
computer security.
|
The "anomalies" to be detected were defined as: suspicious, untypical
entries in a logs of the behaviour of computer network users. The
purpose of this anomaly detection was intrusion detection through
detection of changed user behaviour.
|
|
R&D on detection of "anomalies" in logs of the
behaviour of computer network users. Research into data analysis
methods, such as statistical methods, that can be used in anomaly
detection; followed by design, implementation, documentation, and
testing of a prototype program that executes anomaly detection in
computer logs. (ANSI C, SQL, Sybase, UNIX, Windows NT, MVS.)
|
|
- 1998-2000 (Freelance work, Netherlands)
Analyzing and documenting the logical
functionality of existing programs, to open up the software to the
customer for re-implementation and/or maintenance
|
|
for the following two programs:
1. (for Europe Data Consult, Rotterdam) an old Pascal program
for computation of technical specifications of a electrical power
conductor;
2. (for RIKS, Ministry of Water Management, The Hague,
and EDS, Leidschendam) the DIGIPOL program, a C program for
interpolation of sea- and river depth measurement data.
|
Various other assignments:
|
|
(for Rentmeester Informatisering, Alphen a/d Rijn) conversion of
Windows '9x printer driver software to Windows NT 4.0;
(for EDS Leidschendam) UNIX shell scripting and documentation for a
backup system.
|
|
- 2001 (ED&T, Philips Research, Eindhoven, Netherlands)
Re-implementation of a program for conversion of the
model description of a mixed analog/digital IC.
|
|
The purpose of this program (called "maketiming") was to convert the
model description of a mixed analog/digital IC into an (approximate)
fully digital IC, so that the timing behaviour of the mixed
analog/digital IC could be computed by simulation software
intended for fully digital ICs.
|
|
|
Re-implementation of the program to another internal data structure.
This was necessary to enable the software to handle additional
digital cell library file formats.
Additionally: Documentation and debugging of a few auxiliary
programs used in translation of digital cell libraries to
other file formats. (UNIX, C++)
|
|
- 2002-2004 (im-brain GmbH, Dortmund, Germany)
Design and development of semantic text association software.
|
|
This text association software is a "brain" that can compare texts
in a human-like way, by taking into account the (automatically
generated) "cloud" of associations around each word.
|
|
|
Main task: Design, implementation and testing of the central text
association module used by all im-brain applications.
Additionally: Research into extension of the association
machine to image data, and delivery of prototype systems to compare
logo images and to detect blocks of text in scanned pages.
Delivery of general-purpose hierarchical clustering software, with
graphical visualization of the clustering result. Improvement of
the structure of one of the (server-based) end-user programs.
In all im-brain software, organization of the code into libraries,
and help with making the code more error-free. Writing of software
design and algoritm documentation. (Linux, C++)
|
|
- 2005 (Docuserve, Hamburg, Germany -- for customer Basler Vision
Technologies, Ahrensburg, Germany)
Writing a programmer's guide (API overview) for the
software library of a "smart camera".
|
|
The camera incorporated a general-purpose processor, on which user
programs could be run (typically for image processing).
Delivered with the camera is a software library, which allows
these user programs to control the camera's image capturing and image
transfer hardware.
|
|
|
Main task: Writing an API overview text (programmer's guide)
for this software library. The purpose of this text was to give the
application programmer an overview picture over the library's API, which
the automatically (doxygen) generated "API Reference" could
not give. This task included giving assistance in getting clear
(from the programmers and from the code) which methods/classes
from the library software actually belonged to the exported API.
Additionally: Help with structuring the "API Reference"
generated automatically from comments in the source code; Writing
of explanatory/tutorial texts to go with the set of sample programs
delivered with the camera.
(The software library was a C++ library, using templates
and exceptions. The processor on the camera was running the
Linux operating system.)
|
|
- 2006 (JULIE Lab, University of Jena, Germany)
Java software development for a natural language processing research
group.
|
|
Design and implementation of an infrastructure system used for data
processing and data storage needed for "supervised learning" of language
processing software modules, i.e. adaptation of their parameters to
training samples. This system included a central data repository, a
server for access to this repository, and client programs (for management
of the repository, and for running an external GUI program for entering
the expected outputs to selected training samples).
(Java, TCP/IP, Linux.)
|
|
- Sep 2006 - Dec 2007 (Manu-Dextra GmbH, Nürnberg, Germany -- for customer
Automotive Distance Control Systems GmbH, Lindau/Bodensee, Germany)
Architecture/design documentation,
and development, for the microcontroller software
in a radar sensor device.
|
|
The software on the two microcontrollers (Renesas M32192) in this
radar sensor device computed from the raw radar data the position
and speed of the (nearest) objects in the region swept by the moving
radar beam. The software (ANSI C) was strictly modularized into
separate software components, much like C++ classes. (These modules
largely conformed to the automotive AUTOSAR standard.)
|
|
|
My tasks during this assignment consisted to 75% of software
documentation, and to 25% of sofware development:
Documentation: Software architecture documentation
(description of the application software in the device as a whole,
treating the software components as black boxes), both for the
radar sensor device, and also (as a separate document) for the
general architecture of the suite of basis software components
used on all devices manufactured by the company.
Design documentation for the boot-loader software on both
microcontrollers in the radar sensor device. Design documentation
for one of the application software components in the radar
sensor device, namely the "ACTL" component (Application ConTroL),
responsible for controlling (coordinating) the cyclic data
processing in the device.
Software development: Extension of the "ACTL" software
component, namely making the state machine of the central "operation
mode" of the device configurable at compile time via Excel tables.
These human-readable Excel tables contain VBA code that generates
C source files containing the definition of static C arrays and
structs; this static data constitutes the "program" that controls
the operation of the state machine during run time. Additionally,
writing of Perl scripts for code generation for the software
component operating the CAN bus, and customer adaptation of that
same CAN component.
For the software writen/modified by myself, I delivered extensive
documentation as well.
(C, CAN, OSEK/VDX, Vector/Lauterbach tools, Perl, VBA)
|
|
- Jan-Jun 2008
Gap. This period I intentionally I took off work, to experiment at home
with 8-bit AVR microcontrollers, in order to deepen my understanding
of the hardware aspects and low-level (assembler) programming of
microcontrollers.
|
- Sep 2008 - Jun 2010 (DAKO GmbH, Jena, Germany)
Two tasks:
Design and implementation of a new 3D-kernel for 3D CAD programs
(Main task).
|
|
The company DAKO owns its own 3D CAD (Computer Aided
Design) programs for creation and viewing of catalogs of mechanical
components. As in many 3D CAD programs, the CAD user here
creates complex components step-by-step, by means of intersection,
subtraction, and union operations on simple 3D solids ("Constructive
Solid Geometry", CSG).
|
|
|
My task was to extend the functionality of these CAD programs
so that every solid (3D component) is available not only as a
"recipe" of successive CSG operations (which the software could
already do), but also as a so-called "boundary representation",
namely as the set of connected surface elements (facets) and
edges that the surface of the solid consists of.
To accomplish this, I designed and implemented a completely new 3D
kernel (software library, software component). The most important
task of this 3D kernel is to execute (compute) a CSG operation,
where both the two input solids as well as the output solid are
all represented by their surface elements and edges; and where
the output solid can be fed as input into further CSG operations.
Only four surface element types were needed: plane, cylinder,
cone, and torus.
The 3D kernel software was strictly isolated from OpenGL and from
all visual rendering.
Sub-tasks included: Designing datastructures for the surface
segments and edges; Computation of the intersection curves between
two "arbitrary" surfaces (analytic in simple cases, numeric for
complex surface segments); Writing a tessellator usable for curved
surfaces; Adding a regression tester (indispensable because of the
complexity of the software, and the large quantity of mathematical
special cases).
(C++, OpenGL, MS Visual C++, Windows PC)
|
Programming of a parser and interpreter for a new script language
(2 months).
|
|
In order to allow the manufacturer of the mechanical components to
program the configuration of his catalogs himself, DAKO decided to
develop a new, BASIC-like scripting language.
|
|
|
Assistance with fixing an exact grammar; Documentation of the
grammar; Hand-coding of a "top down" parser (predictive parser)
and of an interpreter. The script language included functions and
local variables. Parser and interpreter implemented as strictly
separate components (and each isolated from the client code by C++
"interfaces").
(C++, MS Visual C++, Windows PC)
|
|
- Oct 2010 - May 2011 (e.sigma Systems GmbH, München, Germany)
Documentation, source code analysis, and maintenance for a
military simulation installation.
|
|
The company e.sigma delivers and services military training
and simulation installations. I worked on location in their
biggest simulation installation, delivered to the German army.
The purpose of this installation is to test guidance systems
used in missiles, i.e. to test whether the sensor and image/data
processing electronics in the missile is able to detect and
track targets successfully. The installation consists of a
spherical projection surface (radius 20 m) that can reflect both
visual and infrared wavelengths; the device under test is placed
in the center of the sphere, mounted on a 3-axis turn table.
Virtual scenarios (landscapes with moving targets) are presented
to the device under test, from a battery of 27 visual projectors
and from three IR laser guns mounted on the inner axis of a
2-axis turn table.
The installation was controlled by around 50 separate PCs,
each with its own special task. These PCs were networked via 3
separate Ethernet networks (TCP/IP). Additionally, a Reflective
Memory network provided "real time" communication between most
of these PCs (apart from the image generators); the PCs thus
interconnected all used Real Time Linux (= RedHat Linux with
real time patch).
|
|
|
The following Documentation was delivered by me:
Documentation about the central program coordinating the
whole installation (logic, data flows, program states, run of
a simulation). A big diagram showing all PCs and all other
units in the installation and how they are interconnected
(purpose: providing insight into the installation as a whole).
Documentation about a few auxiliary programs (for configuration
of the Reflective Memory/DAQ communication; for distribution of
the image data into the installation), and about how to boot up
and power down the installation.
Source code analysis of the source tree for the programs
responsible for the simulation and projection of the visual
and IR targets. These sources had been written for e.sigma by
a third-party company, which had left behind a huge source tree
(2182 subdirectories, 621 makefiles, 12048 C/C++ source files),
scarcely documented and commented, about whose internal operation
e.sigma almost had no knowledge at all. In this source tree,
I identified the obsolete 60% and the compilable 40%, and in
the latter the sources for both programs and for the libraries
used by them. Finally I made sure that the executables built
from these sources were identical to the executables installed
by ECOM (same symbols).
Maintenance: Porting to a new PC with a new Linux version
of existing bash scripts for control of the UPSes and for booting
up/shutting down the installation (about 10 scripts, using lockfiles,
samba, snmpget/set); updated and extended the existing minimal
installation instruction document. Answering questions of the German
army about the installation and about the software. Receiving their
wishes for extensions/changes, and forwarding these to e.sigma. (bash,
C/C++, GNU tools, TCP/IP, SNMP, networked RTLinux and Windows PCs.)
|
|
