3D Computer Vision Engineer — Hands & Pose

Mecka; New York (NY); On-site; USD 141,000 - 162,000; Full time

Job summary

Mecka is seeking a strong, coding-heavy computer vision engineer to own the classical stack. You will manage 3D geometry, temporal filtering, and inverse kinematics to transform model outputs into clean, calibrated spatial data for delivery.

You’ll partner closely with research teams, understand the models, but not train them. Your focus will scale to the entire pipeline, bridging geometry, filtering, and data quality across projects.

Qualifications

• 5+ years of professional software engineering and computer vision.
• Strong production Python; C++ is a plus.
• Solid multi-view geometry: triangulation, PnP, camera calibration, reprojection.
• Temporal filtering and signal processing: low-pass and Kalman-family filters.
• Inverse kinematics and bone-length constraints on articulated structures.
• Some ML familiarity with pose/mesh outputs and failure modes.
• Data-quality mindset; comfortable with noisy real-world data and QA.

Responsibilities

• Own the classical 3D geometry that places reconstructions in space: multi-view geometry, triangulation, PnP, calibration, undistortion, reprojection.
• Design and tune temporal filters to remove jitter without losing real motion and consistency.
• Run inverse Kinematics and enforce physical constraints to keep reconstructed skeletons valid and clean.
• Turn model output into delivered data, including per-frame results, reports, and QA metrics for scale.
• Build visualizers and dashboards to inspect trajectories and failure cases.
• Be the go-to engineer for classical computer-vision problems across the pipeline.
• Provide support for classical CV challenges across the pipeline as they arise.

Skills

Software engineering

Python

C++

Computer vision

Multi-view geometry

Temporal filtering

Inverse kinematics

Data quality mindset

Deep learning familiarity

Tools

Ceres Solver

GTSAM

Job description

Mecka is seeking a strong, coding-heavy computer vision engineer to own the classical stack. You will manage 3D geometry, temporal filtering, and inverse kinematics to transform model outputs into clean, calibrated spatial data for delivery.

You’ll partner closely with research teams, understand the models, but not train them. Your focus will scale to the entire pipeline, bridging geometry, filtering, and data quality across projects.