Member of Technical Staff, Computer Vision (Perception)

Sieve • San Francisco, CA

12d • Onsite

About The Position

Sieve is a multimodal data lab that curates training datasets across video, audio, images, text, and 3D. The company combines large-scale data infrastructure with multimodal understanding techniques to produce high-quality training data for AI applications. It has a team of approximately 30 people and partners with leading AI labs.

As a perception-focused Machine Learning Engineer, you will develop systems that transform egocentric, robotic, and sensor footage into training data for robotics and world-model teams. You will apply computer vision experience from robotics, devices, or autonomous vehicles to large-scale datasets. Work may include recovering camera poses and scene geometry from first-person video, creating hand and object tracking pipelines, and building benchmarks to assess the impact of data on manipulation policies. The role is onsite at Sieve’s San Francisco headquarters, five days per week.

Requirements

• At least 3 years building production perception or computer vision systems for robotics, AR/VR devices, autonomous vehicles, or similar.
• Hands-on experience with SLAM, visual-inertial odometry, multi-view geometry, or 3D reconstruction.
• Strong Python skills and fluency with PyTorch.
• Comfort with multi-sensor data, including camera intrinsics and extrinsics, IMU, depth, and time synchronization.
• Experience training, fine-tuning, or deploying modern deep learning models for vision.
• Strong intuition for evaluation, edge cases, and what makes data useful for training.
• Ability to own a problem from the customer conversation through deployment of a pipeline.
• Interest in Physical AI and the data that will power it.
• Ability to work onsite at the San Francisco headquarters five days per week.

Nice To Haves

• C++ experience.
• Experience with neural scene representations such as NeRFs or Gaussian splatting.
• Background working with egocentric or teleoperation data.
• Publications or open-source work in perception, 3D vision, or robotics.

Responsibilities

• Build perception pipelines for depth, pose, tracking, segmentation, and 3D reconstruction across large video and multi-sensor datasets.
• Apply SLAM, visual odometry, and structure-from-motion to recover camera trajectories and scene geometry at scale.
• Fine-tune vision and vision-language models for embodied tasks, including hand-object interaction, activity understanding, and scene understanding.
• Develop automated quality-assurance and filtering systems to catch calibration errors, synchronization drift, motion blur, and other data defects.
• Create benchmarks and evaluation frameworks to measure the effect of datasets on downstream robotics and world-model performance.
• Work directly with frontier labs and robotics teams to translate data requirements into production pipelines.
• Ship quickly, measure results, and iterate based on real-world data performance.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Education Level: No Education Listed