# Machine Learning Engineer (Egocentric Object Tracking) at Maxinsights

Machine Learning Engineer (Egocentric Object Tracking) at Maxinsights in Santa Clara. On-site, full-time. Posted July 13, 2026.

- Company: Maxinsights
- Location: Santa Clara
- Workplace: On-site
- Job type: Full-time
- Experience: 3+ years
- Posted: 2026-07-13
- Apply: https://jobs.ashbyhq.com/maxinsights/a81aaa2b-431d-4939-adf3-36b8d4c9a157/application
- Source: https://jobmagic.ai/jobs/machine-learning-engineer-egocentric-object-tracking-maxinsights-fge3kh

## Description

### Job Description:

Job DescriptionWe are looking for a Machine Learning Engineer to join our core research and development team, focused on recovering accurate, temporally consistent 6DoF object poses from egocentric (first-person) video.
Understanding how objects move during human demonstrations is essential for robot learning. In this role, you will develop models and pipelines that turn head-mounted and body-mounted camera streams into reliable object trajectories, capturing object position and orientation throughout manipulation. You will tackle real-world challenges including hand occlusion, rapid camera motion, motion blur, object symmetry, and textureless or reflective surfaces.
You will work across the full stack: camera calibration, object modeling, ground-truth collection and annotation, model training and evaluation, and production deployment at scale. This role suits engineers who combine strong foundations in 3D geometry with modern deep learning and enjoy solving measurable perception problems on real-world data.

Responsibilities
- **Build 6DoF object pose estimation and tracking systems** for monocular, stereo, and RGB-D egocentric video, supporting pose initialization, frame-to-frame tracking, and recovery after tracking failures.

- **Develop tracking pipelines for human–object interaction**, addressing partial and full occlusion, rapid object motion, appearance changes, and challenging object geometries and materials.

- **Combine geometric and learning-based methods**, including visual correspondences, point tracking, segmentation, pose refinement, and temporal optimization.

- **Own camera geometry and coordinate-frame consistency**, including intrinsic/extrinsic calibration, wide-FOV camera models, metric-scale reconstruction, and transformations between object, camera, and world coordinates.

- **Improve trajectory accuracy and temporal consistency**, distinguishing camera motion from object motion and reducing drift through multi-view constraints, robust estimation, and sequence-level refinement.

- **Build the ground-truth and evaluation loop**, including object reference assets, pose annotation tools, quality checks, and evaluation of pose accuracy, tracking robustness, and recovery performance.

- **Ship scalable training and inference pipelines**, collaborating with engineering and robotics teams to deliver object motion data for robot learning.

- Contribute to technical design, code quality, and the long-term direction of the company’s perception stack.

Required Qualifications
- Bachelor’s, Master’s, or PhD in Computer Science, Machine Learning, Computer Vision, Robotics, or a related technical field, or equivalent practical experience.

- 3+ years of experience developing computer vision or machine learning systems.

- **Demonstrated hands-on experience with 6DoF object pose estimation or rigid-object pose tracking from video**, including implementing, training, or adapting methods and evaluating them on real-world data.

- **Strong foundations in 3D geometry**, including camera projection and calibration, rigid-body transformations, rotation representations, coordinate systems, and geometric optimization.

- Practical experience with relevant techniques such as **2D–3D or 3D–3D correspondence estimation, PnP, RANSAC, point-cloud registration, ICP, or render-and-compare pose refinement**.

- Understanding of temporal tracking challenges, including pose initialization, drift, occlusion handling, tracking failure detection, and reinitialization.

- Strong proficiency in Python and PyTorch or an equivalent deep learning framework, with experience building reproducible training and evaluation pipelines.

- Ability to diagnose perception failures, design meaningful experiments, and translate research methods into reliable engineering systems.

Preferred Qualifications
- **Direct experience with egocentric object tracking or hand–object interaction**, particularly under heavy occlusion and rapid camera motion.

- Experience with **model-based and model-free object pose estimation**, using CAD models, scanned meshes, reference images, or reconstructed object representations.

- Experience combining object tracking with **camera pose estimation, SLAM/VIO, depth estimation, or 3D reconstruction** to recover object trajectories in a consistent world coordinate frame.

- Familiarity with object detection and segmentation, dense visual matching, point tracking, or scene flow, and their application to 6DoF pose tracking.

- Experience with multi-camera capture, RGB-D sensors, time synchronization, calibration, or ground-truth systems for object motion.

- Experience with synthetic data generation, differentiable rendering, domain randomization, or sim-to-real transfer.

- Experience building annotation tools, confidence-aware quality checks, or large-scale video processing pipelines.

- Publications at leading venues such as CVPR, ICCV, ECCV, NeurIPS, SIGGRAPH, or 3DV; relevant open-source contributions; or demonstrated impact in applied computer vision systems.

Default Benefits:
- Health insurance

- Vision care

- Dental coverage

- 401(k)

- Paid holidays

- PTO (Paid Time Off)

- Sick leave