Video Compression Engineer

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 reduce the bandwidth needed to send robot video to shared compute while preserving the information robot models depend on. You'll own the path from ISP output through encoding and work with wireless and evaluation engineers to adapt compression to network conditions and task requirements.

What you’ll do

• Build low-latency encoding pipelines on embedded hardware, tuning bitrate, quantization, frame structure, resolution, and frame rate for robot workloads.
• Develop adaptive compression policies using link capacity, queue age, scene motion, and task feedback, with stable behavior as network conditions change.
• Work with ISP engineers on noise, sharpening, exposure, and temporal processing to improve compressibility without removing task-relevant detail.
• Measure rate-distortion curves alongside VIO, perception, and robot-policy outcomes; identify where additional compression causes a meaningful loss of capability.
• Optimize buffer management, memory transfers, hardware encoder use, and decoder handoff so compression saves bandwidth without adding avoidable delay.
• Test packet loss, reference-frame recovery, abrupt scene changes, and encoder reconfiguration; maintain reproducible benchmarks across cameras, codecs, and hardware versions.

What we are looking for

• Experience implementing or tuning video codecs, embedded media systems, or real-time image-processing pipelines.
• Strong understanding of H.264/H.265, prediction structures, rate control, quantization, and the tradeoffs between quality, bitrate, and latency.
• Proficiency in C/C++ on Linux and experience with frameworks such as V4L2, GStreamer, or FFmpeg.
• Ability to profile memory movement, CPU load, and hardware codec behavior and resolve frame drops or unexpected buffering.
• Experience designing controlled video experiments and interpreting both image-quality metrics and downstream application results.

Nice to Have

• Experience with learned compression, region-of-interest coding, or compression optimized for machine vision.
• Knowledge of ISP tuning, RAW/Bayer data, hardware codecs, or zero-copy camera pipelines.
• Work on real-time transport, congestion feedback, or robotic-policy evaluation under visual degradation.