Senior Perception Engineer

Atoms•San Francisco, CA

22h•$208,000 - $245,000•Onsite

About The Position

We are seeking a Perception Engineer to develop the systems that enable our autonomous machines to understand and interact with the physical world. You will work on perception problems at the intersection of computer vision, machine learning, robotics, and autonomous systems, building models and systems that transform real-world sensor data into reliable representations of the environment.

Requirements

3-5 years of relevant experience working with perception systems and computer vision & machine learning models.
Strong software engineering skills in Python and/or C++.
Experience developing perception systems for autonomous vehicles, robotics, drones, industrial automation, or other embodied systems.
Experience with camera, LiDAR, radar, IMU, or other robotic sensor modalities.
Experience with multi-modal or multi-sensor architectures.
Knowledge of 3D geometry, coordinate transformations, calibration, localization, or state estimation.
Experience deploying ML models into real-time or safety-critical systems.
Experience optimizing models for latency, compute, memory, or power constraints.
Experience working with large-scale perception datasets and training infrastructure.
Experience diagnosing long-tail failures and improving production perception systems through data-driven iteration.
Experience taking research ideas from experimentation through deployment on physical systems.

Nice To Haves

Publications or significant research contributions in computer vision, machine learning, robotics, or related fields are a plus, but not required.

Responsibilities

Design, train, and deploy machine learning models for perception in autonomous systems.
Develop algorithms for object detection, classification, segmentation, tracking, and scene understanding.
Build systems for 3D perception and geometric reasoning using cameras, LiDAR, radar, or other sensors.
Develop approaches for multi-sensor and multi-modal fusion, combining complementary sensor information into robust representations of the environment.
Build temporal models that reason about objects, motion, trajectories, and changes in the environment over time.
Improve perception performance across challenging real-world conditions, including occlusion, sensor noise, changing environments, and edge cases.
Develop data, training, evaluation, and experimentation pipelines that accelerate perception development.
Define metrics and evaluation methodologies that connect offline model performance to real-world system behavior.
Optimize perception models and systems for deployment on resource-constrained or latency-sensitive robotic platforms.
Investigate failures observed in deployed systems and translate those findings into improvements across models, data, and architecture.
Collaborate closely with AI Research, Robotics, ML Infrastructure, Software, Hardware, and Systems Engineering teams.