AI Engineer, Computer Vision

About Mytra

We’re creating an entirely new way to solve the most ubiquitous problem in industry - moving and storing material. We’re applying robotics and distributed software to create a new class of product for this $1T market. We’re focused on the supply chain industry first. The industry is in a massive bind with the continued growth of e-commerce, sharp rise in costs, and supply chain disruptions. What has been a “sleepy” industry for decades is now at the epicenter of sustaining the global economy.

Role Overview

Mytra's AI team seeks pioneering engineers to architect the neural backbone of our distributed robotic intelligence system. As a Computer Vision AI Engineer, you'll push the boundaries of multi-agent perception by developing novel approaches for real-time scene understanding, geometric deep learning, and distributed spatial cognition. You'll tackle fascinating challenges like dynamic SLAM in unstructured environments, multi-view 3D scene reconstruction, collective mapping, and real-time object detection/tracking with resource-constrained edge deployment. Working at the intersection of computer vision and robotics, you'll implement state-of-the-art architectures for depth estimation, pose estimation, and semantic segmentation while collaborating with a multidisciplinary team to bridge the gap between research and production-grade systems.

Example Projects

• Train a depth network for monocular camera and time-of-flight sensor data.
• Implement a computer vision model from a newly published paper.
• Optimize a model architecture for size, speed and accuracy given edge computing resource constraints and integrate into a realtime production perception system.
• Train models despite limited data using techniques such as synthetic data generation, data augmentation, and domain adaption.
• Create an occupancy map for localization and collision avoidance from vision depth maps and other learned features.
• Develop automated processes and benchmarking strategies for tracking improvements when releasing and deploying new models to the robot fleet.
• Leverage new open source tools or models to speed up the machine learning pipelines (such as using Segment Anything Model (SAM) to auto-label images).
• Develop reliable and scalable methods for camera calibration.
  
  

The Ideal Candidate

• You are a computer vision enthusiast, theoretical yet pragmatic, curious yet focused, always learning yet humble, and you are motivated to see your work deployed at scale, solving real world problems.
• You’re very comfortable writing Python and have experience with PyTorch and training deep neural networks.
• You have an owner's mindset. You are a doer.
• You are excited about working on new projects, growing together, and shaping the future of the company.
• You’ve played with different types of sensors and are familiar with robot software frameworks.
• Uncertainty isn’t scary, and you’re willing to realize mistakes and pivot.
• You’re familiar with simulators such as Isaac Sim, OpenAI Gym, MuJoCo, Unity, or other video game environments.
• You can convey complex ideas and technical concepts to a diverse audience.
• Openness, inquisitiveness, and constant ambition are important values.
• You enjoy pair programming and taking time to help others.
  
  

A Final Note

If this is your dream role, but you can’t take on every example project in its totality, we encourage you to apply. We work closely and collaboratively at Mytra - no one takes on projects alone. We seek people who are eager to work together, learn new things, and bring unique perspectives.

• Mytra is not currently utilizing third-party agencies.
  
  

The Pay Range For This Role Is

140,000 - 180,000 USD per year(South San Francisco)