Algorithmik, SW-Architektur, C++. Schwerpunkt auf Bildverarbeitung und numerische Simulationen mit GPU-Beschleunigung (CUDA/OpenCL).
Aktualisiert am 03.01.2025
Profilbild
Profil
Freiberufler / Selbstständiger
Remote-Arbeit
Verfügbar ab: 07.01.2025
Verfügbar zu: 100%
davon vor Ort: 0%
CUDA
Algorithmenentwicklung
C++
C++17
Softwarearchitektur
Bildverarbeitung
Chinese
level A1
English
2. native language
French
basics
German
native language
Mandarin
level A1

Einsatzorte

Einsatzorte

München (+500km)
Deutschland, Schweiz, Österreich
möglich

Projekte

Projekte

Software Developer
  • Implemented new features into GUI-Tools
  • Supported working students and new employees

Developed, implemented and applied for a patent

  • For a new, lossless raw image compression format optimized for real-time GPU raw image compression and decompression
  • Achieved 15% smaller footprint and faster processing than competing format

Optimized and improved quality of existing image-processing pipeline

  • Achieved over 2x aggregate performance in CUDA by overhauling every kernel
  • Also helped enable multi-GPU support and asynchronous processing

Implemented and contributed to new algorithms

  • Contributed to a machine-learning based dead pixel detector
  • Contributed to noise reduction and demosaicing algorithms
  • Developed a nvidia tegra CUDA based internal test station
  • Helped make new algorithms real-time capable
Arnold & Richter Cine Technik GmbH (ARRI)
Working Student
  • Faculty for Experimental Physics with Cosmic Particles (ECP)
  • Implemented features necessary for new detectors into simulation framework

Ported CPU- based analysis and Reconstruction tools to CUDA and OpenCL

  • Achieved significant improvements in processing speed by GPU processing and by restructuring programs
  • Reduced processing time of an in-use algorithm from hours to seconds while enabling increased precision

Contributed to new reconstruction and analysis tools

  • Developed awareness of the deteriorating effect of common pseudo-random number generators on simulation-generated control data
  • Helped improve understanding of the effect of machine precision on algorithms
Technical University of Munich (TUM)
Tutor

Self-employed through personal recommendation. Concurrently to university studies.

  • Tutored a total of over 120 students in Math, Physics, Chemistry and Computer Science in all grades and schools
  • Developed communication and motivation skills
  • Experienced quite a few surprises

Aus- und Weiterbildung

Aus- und Weiterbildung

Computer Science
Technical University of Munich (TUM)

Quit due to excellent job offer

Bachelor of Science
Technical University of Munich (TUM)

ERASMUS semester in Trondheim, Norway in winter semester 2016/17

Kompetenzen

Kompetenzen

Top-Skills

CUDA Algorithmenentwicklung C++ C++17 Softwarearchitektur Bildverarbeitung

Produkte / Standards / Erfahrungen / Methoden

CUDA:

  • Self-Learned since first version on fermi graphics cards
  • Hobby projects and professional application
  • Deep knowledge of graphics architectures and
  • performance impacts/ optimization Strategies

C/C++:

  • Self-Learned at age 18
  • Improved through experience on IceCube and ARRIRAW projects

OpenCL:

  • Learned by experience on IceCube and ARRIRAW projects

Java:

  • First self-Learned object oriented programming language
  • Started form age 12 with hobby projects

Python:

  • Learned in university
  • Mainly through machine and deep learning

Git, Subversion:

  • Learned by experience on IceCube and ARRIRAW projects

Einsatzorte

München (+500km)
Deutschland, Schweiz, Österreich
möglich

Projekte

Software Developer
  • Implemented new features into GUI-Tools
  • Supported working students and new employees

Developed, implemented and applied for a patent

  • For a new, lossless raw image compression format optimized for real-time GPU raw image compression and decompression
  • Achieved 15% smaller footprint and faster processing than competing format

Optimized and improved quality of existing image-processing pipeline

  • Achieved over 2x aggregate performance in CUDA by overhauling every kernel
  • Also helped enable multi-GPU support and asynchronous processing

Implemented and contributed to new algorithms

  • Contributed to a machine-learning based dead pixel detector
  • Contributed to noise reduction and demosaicing algorithms
  • Developed a nvidia tegra CUDA based internal test station
  • Helped make new algorithms real-time capable
Arnold & Richter Cine Technik GmbH (ARRI)
Working Student
  • Faculty for Experimental Physics with Cosmic Particles (ECP)
  • Implemented features necessary for new detectors into simulation framework

Ported CPU- based analysis and Reconstruction tools to CUDA and OpenCL

  • Achieved significant improvements in processing speed by GPU processing and by restructuring programs
  • Reduced processing time of an in-use algorithm from hours to seconds while enabling increased precision

Contributed to new reconstruction and analysis tools

  • Developed awareness of the deteriorating effect of common pseudo-random number generators on simulation-generated control data
  • Helped improve understanding of the effect of machine precision on algorithms
Technical University of Munich (TUM)
Tutor

Self-employed through personal recommendation. Concurrently to university studies.

  • Tutored a total of over 120 students in Math, Physics, Chemistry and Computer Science in all grades and schools
  • Developed communication and motivation skills
  • Experienced quite a few surprises

Aus- und Weiterbildung

Computer Science
Technical University of Munich (TUM)

Quit due to excellent job offer

Bachelor of Science
Technical University of Munich (TUM)

ERASMUS semester in Trondheim, Norway in winter semester 2016/17

Kompetenzen

Top-Skills

CUDA Algorithmenentwicklung C++ C++17 Softwarearchitektur Bildverarbeitung

Produkte / Standards / Erfahrungen / Methoden

CUDA:

  • Self-Learned since first version on fermi graphics cards
  • Hobby projects and professional application
  • Deep knowledge of graphics architectures and
  • performance impacts/ optimization Strategies

C/C++:

  • Self-Learned at age 18
  • Improved through experience on IceCube and ARRIRAW projects

OpenCL:

  • Learned by experience on IceCube and ARRIRAW projects

Java:

  • First self-Learned object oriented programming language
  • Started form age 12 with hobby projects

Python:

  • Learned in university
  • Mainly through machine and deep learning

Git, Subversion:

  • Learned by experience on IceCube and ARRIRAW projects

Vertrauen Sie auf Randstad

Im Bereich Freelancing
Im Bereich Arbeitnehmerüberlassung / Personalvermittlung

Fragen?

Rufen Sie uns an +49 89 500316-300 oder schreiben Sie uns:

Das Freelancer-Portal

Direktester geht's nicht! Ganz einfach Freelancer finden und direkt Kontakt aufnehmen.