Image Processing Engineer, Video Stabilization & Motion Estimation

About the company

Efference provides high-performance perception and compute infrastructure for training and deploying robot models in the cloud. Our vision is to make robotic deployments scalable with shared compute.

We approach the problem by owning the entire path from the image sensor to the datacenter GPU. We shrink the load at the sensor, fix the uplink so robot traffic is fast and reliable, and clock synchronize the system so observations arrive time-aligned, ready for inference.

About the role

You'll develop motion estimation and video stabilization for cameras carried by robots that walk, turn, vibrate, and interact with their surroundings. You'll make video useful for perception and teleoperation while accounting for the geometry, timing, and field of view that downstream systems require.

What you’ll do

• Develop image-based, gyro-based, and hybrid motion estimation and stabilization algorithms for wide-angle, fisheye, and conventional camera systems.
• Build camera/IMU calibration and alignment methods with the VIO and embedded teams, identifying bias, timing offsets, and mechanical changes that affect compensation.
• Implement warping and reprojection that handle rolling-shutter distortion, rapid rotations, vibration, and motion blur within the device's compute budget.
• Evaluate the tradeoffs between residual motion, crop, geometric distortion, temporal filtering, and added latency on representative robot trajectories.
• Preserve the calibration and transform metadata needed by downstream consumers; test when VIO or policy inference should use original rather than stabilized frames.
• Prototype in Python and ship optimized implementations on embedded processors, with regression tests for motion accuracy, visual artifacts, compression, and task performance.

What we are looking for

• Hands-on experience developing video stabilization, optical flow, camera motion estimation, or closely related geometric vision algorithms.
• Strong understanding of camera models, projective geometry, image warping, and the limits of motion estimates in difficult scenes.
• Experience with IMUs, sensor fusion, and spatial and temporal calibration between camera and inertial measurements.
• Strong C++ and Python skills and experience turning algorithm prototypes into real-time implementations.
• Ability to evaluate long video sequences and explain failure cases using measurements, image evidence, and a clear understanding of the camera pipeline.

Nice to Have

• Experience with robotics, action cameras, drones, or AR/VR tracking systems.
• Work with learned optical flow, neural motion estimation, or hardware-accelerated image warping.
• Knowledge of VIO/SLAM, ISP tuning, global-shutter sensors, or embedded GPU/DSP/NPU deployment.