Machine Learning Vision Engineer

About the role
We're looking for a Machine Learning Vision Engineer to help our R&D team push the boundaries of video analytics — from multi-view geometry and 3D reconstruction to multiple object tracking and facial analysis. You'll work at the core of Aurora, our own cloud-managed stereo sensor, which combines onboard shopper detection and tracking with WiFi, Bluetooth, and HD video recording.
You'll do well here if you have a strong grip on multi-view geometry, an inventive eye for new techniques, and a genuine drive to scale computer vision into large, real-world installations. This isn't research in a vacuum — the algorithms you build run in stores around the world.
Because RetailNext builds its own hardware and software end to end, you'll get a rare shot at owning problems from theory through production deployment.
Reports to: Head of Technology
Location: Taiwan, Remote
What You'll Do
Algorithm Research & Development
Lead the development of new computer vision algorithms for shopper detection, tracking, and analytics
Improve the performance and efficiency of existing algorithms already running in production
Apply multi-view geometry techniques to solve image registration, mosaicking, and 3D reconstruction problems
Prototype new approaches quickly in Python or MATLAB before hardening them for production
Stay close to the research space so RetailNext's techniques keep pace with — and get ahead of — the field
Multi-Sensor Systems Integration
Build robust software that integrates multiple sensors and tracking systems, including Aurora's stereo cameras, WiFi, and Bluetooth
Design for real-world conditions — variable lighting, store layouts, and hardware constraints — not just clean lab data
Work across the video, WiFi, and Bluetooth analytics pipelines to keep detection and tracking accurate and consistent
Applied Engineering & Delivery
Design, implement, validate, and release applications and capabilities in C++ and Python
Take algorithms from prototype to production-ready code that runs reliably at scale
Validate new techniques against real deployments across large, global installations
Partner with the rest of the R&D team to review, test, and ship new capabilities
What We're Looking For
Required
Master's degree in Computer Science, Computer Vision, or Machine Learning
Strong programming skills in C/C++
Rapid prototyping experience in Python or MATLAB
Strong understanding of multi-view geometry
Experience designing, implementing, and releasing computer vision applications
Valued but Not Required
Experience with multiple object tracking or facial analysis
Experience integrating multi-sensor systems (video, WiFi, Bluetooth)
Background applying computer vision at production scale, not just in research
Experience with embedded or onboard processing on custom hardware
What We Offer
Work on novel computer vision problems most teams never get near — multi-view geometry, 3D reconstruction, and multiple object tracking at production scale
Help shape the algorithms behind Aurora, our own cloud-managed stereo sensor, from R&D through installations running around the world
Real ownership over new algorithm development, not just maintenance — your work directly drives what ships next
Monthly Recharge Day
Holiday Exchange Program
Best Self Benefit Program
Medical Insurance
Work from Anywhere up to 90 days

Work arrangement
Yes