Senior Researcher - Acceleration of AI models - LifeworQ Jobs GmbH
  • N/A, Other, Canada
  • via LifeworQ Jobs GmbH
-
Job Description

Our team has an immediate permanentopening for a Senior Machine Learning Research Engineer.
Responsibilities:

  • Track the trend of AI theory and technology development in the world and generate research report and proposals for promoting Ascend system accordingly.
  • Lead or participate in research of algorithms in accelerating the training of the market-driven AI models (CV/NLP/GNN/...), reaching/exceeding the state of the art accuracy, and develop a proof of concept of the algorithms. Those algorithms include but are not limited to the following: optimizers, loss functions, new model architecture, mix precision, model compression, learning technologies (e.g., meta-learning), etc.
  • Publish relevant high-quality AI research papers when necessary and approved, and attend conferences for increasing public awareness of Huawei’s Ascend products; file high-value patents on critical algorithms/processes that are of potential business gain.
  • Team up with other departments/teams from Huawei’s global research centers for collaboration.
  • Assist the team lead on theplanning of projects and definition of technology/products development road map.

What you’ll bring to the team:

  • Master/PhD in Computer Science, Math/Statistics, focusing on AI & Deep Learning with solid publication records.
  • 2+ years working experience in optimizing performance of training deep learning models and/or their applications to CV/NLP/GNN domains.
  • Solid skills in programming in Tensorflow/Keras/PyTorch/MXNet.
  • Hands-on skills in C++/Python programming.
  • Excellent documentation skills in writing internal reports and/or publishing research papers.
  • Excellent communication skills in internal and external presentation.
  • Working knowledge of AI accelerators or the full stack of AI acceleration system is anasset.
  • Strong math background in optimization (e.g., gradient descending) is an asset.

#LI-SZ2

#J-18808-Ljbffr

;