Researcher– High Performance and Parallelism in Data Science (M/F)

Référence : ITIS-2020-015

Type: Chercheur
Type de contrat: CDD
Durée: 24 months
Lieu de travail: Belval

Contexte

 

Your work environment

The Luxembourg Institute of Science and Technology (LIST) is a Research and Technology Organization (RTO) active in the fields of materials, environment and IT. By transforming scientific knowledge into technologies, smart data and tools, LIST empowers citizens in their choices, public authorities in their decisions and businesses in their strategies.

https://www.list.lu/

The Luxembourg Institute of Science and Technology (LIST) in partnership with Goodyear is driving significant research and development focused on responding to innovation challenges in tire-related applications and business intelligence. All projects and positions in this partnership are co-supervised by LIST senior researchers and our industrial partner professionals who all have a large experience in supervision and project management.

You will be part of the LIST IT for Innovative Services department

The “IT for Innovative Services” (ITIS) department of LIST is home to 110 IT scientists and engineers performing research and innovation activities in computer and data science topics applied, in particular, to the design, engineering and assessment of novel approaches to information systems and services. These activities are supported by a Data Analytics/AI technological platform including HPC, Visualisation Wall and a Cognitive pillar.

The candidate will join the Data Science and Explainable Analytics Unit whose mission is to imagine, design, implement and assess novel approaches based on data sciences and interactive technologies to support problem-solving and decision-making in an explainable manner with human in the loop. The Unit activities include research projects as well as more applied activities carried out with private and public partners.

 

Description

 

Researchers in the Goodyear Data Science department develop innovative tools and methods for simulating, modelling and analysing complex processes throughout the company. This often yields computationally heavy tasks (e.g. hyper-parameter tuning), and effective code optimization and parallelization then becomes crucial.

We are looking for a highly motivated researcher to work on:

  • Defining a taxonomy of data science programs and algorithms that may benefit from (CPU and GPU) parallelization facilities
  • Identifying relevant algorithmic and programming design patterns
  • Translating bottlenecks observed in profiling tools to design patterns
  • Designing and implementing software to automate code transformation and favour adoption by data science practitioners

Effectively parallelizing legacy (C or Python) code and algorithms does not always come easy. In addition, even in the recourse to frameworks such as Tensorflow and PyTorch, several pitfalls and bottlenecks may prevent maximal gains from GPU hardware usage. Existing profiling tools (e.g. Tensorboard Profiler, Nvidia Nsight) provide insights regarding these bottlenecks, but interpreting their results and translating the latter into program modifications often require advanced programming skills. The goal of the researcher will be to advance the state of the art in advanced profiling and code transformation and contribute to bridging this gap.

 


Profil

 

Educational background

  • PhD degree in Computer Sciences or Applied Mathematics

Technical skills

  • Expertise in a research field overlapping data science, software engineering, and / or high-performance computing
  • Demonstrated experience in C/C++ programming in demanding contexts regarding computational performance (e.g. CUDA, OpenMP, OpenACC)
  • Experience in using C/C++/Python profiling tools for the identification of computational bottlenecks, in the context of CPU or GPU-based computation
  • Proven experience in Python programming and using frameworks and libraries such as Numpy, scikit-learn, Keras or Tensorflow.
  • Experience in language engineering and meta-programming is a plus
  • Knowledge of classical machine learning techniques
  • Strong interest in engaging in innovative value-creation activities with industrial partners
  • Excellent written/verbal communication skills with the proven ability to effectively communicate across different levels of the organization and with stakeholders/project partners
  • Result oriented attitude
  • Willingness to work collegially in a multidisciplinary and multicultural environment

Language skills

  • Excellent communication skills (oral, written, presentation)
  • Proficient written and spoken English is mandatory

 

Contact

Candidates interested in the above position can apply online on our website www.list.lu
The application file should include:

  • A resume
  • A motivation letter
  • The names of three referees
  • A list of publications

 

Partager cette page :

ITIS-2020-015


 

Postuler en ligne

Contact

Dr Ir Benoît OTJACQUES
Dr Ir Benoît OTJACQUES

Dr Pierrick BRUNEAU
Dr Pierrick BRUNEAU

Dr Thomas TAMISIER
Dr Thomas TAMISIER

Pas de photo
Aldjia AFIRI