Posted in

Staff Deep Learning Engineer

Staff Deep Learning Engineer

CompanyHayden AI
LocationSan Francisco, CA, USA
Salary$Not Provided – $Not Provided
TypeFull-Time
DegreesMaster’s, PhD
Experience LevelSenior, Expert or higher

Requirements

  • Ph.D. or Master’s in Robotics, Machine Learning, Computer Science, Electrical Engineering, or a related field.
  • Expertise in PyTorch or TensorFlow (one mandatory, familiarity with both a plus).
  • OpenCV for computer vision.
  • TensorRT for deployment optimization.
  • Strong Python programming and software design with experience in Pandas.
  • Experience deploying DL models to run on real-world, resource-constrained systems.
  • Demonstrated proficiency in data science and traditional machine learning (SVMs, Random Forests).
  • Experience in automated data annotation for computer vision.
  • Training multi-task and semi-supervised deep learning models for video data.

Responsibilities

  • Drive the entire perception system development life cycle, from problem definition to deployment and ongoing improvement.
  • Actively contribute to the development and refinement of the perception system in a hands-on manner.
  • Develop robust computer vision algorithms for object detection, tracking, semantic segmentation, and classification.
  • Design and train deep learning models for complex urban scene perception and real-time analysis.
  • Collaborate with cross-functional teams (cloud/device) for seamless integration and monitoring of perception models.
  • Analyze data to identify performance bottlenecks and opportunities for enhancing the perception system.
  • Automate improvement cycles of deep learning models used within the perception system.
  • Communicate technical findings and insights effectively to stakeholders across the company to drive performance improvements.
  • Utilize data visualization tools to present complex information clearly for informed decision-making.

Preferred Qualifications

  • Familiarity with designing multi-modal deep learning models incorporating temporal context and geometrical constraints is a plus.