Senior Software Engineer, Perception, Machine Learning/computer Vision

Nuro Nuro · Robotics · CA · Autonomy

Senior Software Engineer focused on perception ML/CV for Nuro's autonomous driving platform. The role involves applying deep learning to challenges like detection, sensor fusion, tracking, and human intent understanding. Responsibilities include all stages from proof of concept to onboard deployment and performance monitoring, with a focus on training models for massive data utilization while maintaining low onboard latency. Experience with real-time deployment and optimizing accuracy vs. compute tradeoff is a bonus.

What you'd actually do

  1. You will be involved in all stages of problem solving, including initial proof of concept, model iteration, onboard deployment, and on-road performance monitoring and troubleshooting.
  2. You will develop techniques and/or processes that allow us to train models that are able to utilize massive scale of data, but still keep the model onboard latency in check so that it is deployable.
  3. You will have the opportunity to take a novel problem, assume the ownership, and just go deep on it.

Skills

Required

  • Recent experience in applying state-of-the-art deep learning techniques to solving autonomous driving problems, such as modern object detection models, sensor fusion, and multiple task learning.
  • Strong ML fundamentals and has demonstrated experience building and expanding model architectures
  • Strong SWE fundamentals and has demonstrated SWE experience in both high and low-level languages (Python and C++)

Nice to have

  • Experience in deploying models in real time environments.
  • Optimizing architectures to explore the accuracy vs. compute tradeoff
  • Experience working with productizing ML models and the infrastructure supporting model development (data, large scale training, eval, etc)

What the JD emphasized

  • deploying models in real time environments
  • Optimizing architectures to explore the accuracy vs. compute tradeoff

Other signals

  • applying state-of-the-art deep learning techniques to solving autonomous driving problems
  • develop techniques and/or processes that allow us to train models that are able to utilize massive scale of data, but still keep the model onboard latency in check so that it is deployable
  • deploying models in real time environments
  • Optimizing architectures to explore the accuracy vs. compute tradeoff