Job description
As a Computer Vision Engineer at Exodigo, you will work on problems where imagery meets the physical world. Our systems observe real above-ground environments and have to produce results that are accurate in real-world coordinates — which means the work lives in the transition between 2D imagery and 3D space: reconstruction, localization, depth, registration, and geometric reasoning about what the sensors actually saw.
The core of the role is solid computer vision engineering. Detection, segmentation, tracking, and the unglamorous work of making these tasks accurate and robust on messy real-world data are the backbone of what we build. Our stack is modern and constantly evolving — we adopt new methods as the field produces them, and we care more about whether something works on our data than about how new it is. The loop here is short — what you build reaches our mapping experts and makes an impact!
Key Responsibilities
Own the development of vision algorithms end to end — from framing the problem using real data, through model design and iteration, to results that meet our accuracy requirements in production.
Develop computer vision algorithms that turn imagery and multi-sensor data into a reliable interpretation of real environments — the scene understanding that our mapping outputs depend on.
Work on the 2D-to-3D problem: geometric reasoning, reconstruction, localization, depth, and registration of visual data to real-world coordinates.
Rapidly prototype, benchmark, and refine models against our own data and accuracy requirements.
Work closely with our Mapping and Engineering experts to validate real-world robustness, and with the Software team to turn prototypes into production-grade systems.
Requirements
4+ years of industry experience developing computer vision and machine learning algorithms. Academic research experience counts in addition to this, not instead of it.
Strong spatial intuition and hands-on experience with the transition between 2D and 3D. This is a core requirement — we expect you to have worked on problems such as navigation, mapping, 3D reconstruction, photogrammetry, localization, depth estimation, or something similar in spirit.
Strong understanding of modern deep learning architectures, including transformer-based and multimodal models — how they are built and where each one fits, even if your own hands-on experience has been elsewhere.
Strong proficiency in Python and the standard scientific and computer vision stack: NumPy, SciPy, OpenCV, and PyTorch.
B.Sc. in Electrical Engineering, Computer Engineering, Computer Science, or a similar field.
Preferred Qualifications
M.Sc. or Ph.D. in a relevant field.
Experience solving problems that combine text and imagery, such as document analysis.
Experience with point-cloud analysis or LiDAR data.
Experience in aerial, orthophoto, or satellite image analysis, or in geolocation problems.
Familiarity with cloud development environments, particularly AWS.
Learn more
