Open Positions

Deep Learning Engineer Computer Vision on Edge

As a versatile computer vision engineer you will design and implement an integrated distributed software solution for multi-camera vision processing on edge devices using Deep Learning methods. You are familiar with methods to transform RGBD data to 3D point cloud representations, camera calibration, perspective projections, 3D modeling/rendering tools and you have proven distributed computing skills using containerised applications for real-time video stream processing.

Your environment will be a small startup with an ambitious vision. This means that you should have a proactive personality with a can-do mentality and you are able to work in a super-agile team. You are not afraid of a little bit of chaos at times, because you have a way of finding your focus on the essential parts that will lead to our short-term and long-term goals. We don’t do micro-management, but we do work together as one strong technical team to tackle the problems we encounter.

Our main goal is to build advanced scene interpretation solutions to build a city-wide situational awareness network. We are involved in the Surv-AI-llance project, which will be part of a larger application suite, called City Scenes. In short, our solution will extend and integrate with existing city camera networks to produce real-time insights to support smart cities.

Requirements

Startup mentality:

  • we’re building robust proof of concepts, with urgency
  • you have a firm solution-oriented mindset: nothing is impossible until proven otherwise

Ph.D. in computer vision or M.Sc. with greater than 5 years’ experience in computer vision on the topic of 3D scene analysis and geometric interpretation

You have experience with video stream processing and common computer vision libraries

  • experience with deepstream/gstreamer is a plus
  • experience with OpenGL or Vulkan is a plus
  • experience with OpenCV is a plus

You are comfortable with C/C++ coding, embedded systems and multi-threaded programming

You are proficient in writing Python applications that are robust and production ready

You are fluent with modern Deep Learning tools and you have hands-on experience with PyTorch or TensorFlow (we mainly use PyTorch)

  • you will work with models for object detection, tracking, semantic segmentation, depth estimation, geometric interpretation, action recognition, full 3D scene understanding, multi-sensor fusion
  • designing new models will not be your main focus, but you may get involved when necessary
  • experience with TensorRT is a plus
  • some example reference code you should feel comfortable with: https://github.com/orgs/NVIDIA-AI-IOT

Affinity for or experience with 3D engines (such as Unreal Engine/Unity/Blender) is a plus

  • we believe synthetic data generation will improve our models

You are a hands-on person:

  • you quickly familiarize with existing codebases and you know how to get things running
  • you know how to connect the dots to tackle complex problems
  • working with linux and distributed software makes you feel comfortable
  • edge devices are your toys, we love the NVIDIA Jetson family
  • you are familiar with building and deploying containerised applications (e.g. docker), experience with Kubernetes/Helm is a plus
  • you rely on scientific literature for selecting promising methods

We maintain a goal-oriented approach based on independent work with team support:

  • you know how to find your focus without losing yourself into the details
  • you document everything that matters
  • you are open minded and comfortable in sharing ideas
  • good work ethics: you have a good team spirit, you easily collaborate with colleagues, mutual respect is a golden rule

We are based in Mechelen (Belgium) and our main language is English, speaking Dutch is an added value

  • our location implies you should live at a commute-friendly distance from Mechelen
  • work from home is still the default, but we now work together once a week in Mechelen

Last point, if you read this far and are interested in the job, but you don’t fill some of the above bullets, don’t hesitate to contact us and explain which bullets you don’t match and how you would overcome them.

Expected start date will be June/July 2022, but can be discussed.

We look forward to receive your application

Apply for this job by sending an email to jobs@atomicbits.ai and attach your CV and link to your LinkedIn profile or any other online references (e.g. your github projects) that illustrate your abilities.