Computer Vision / Hand-Tracking Engineer - PT - US-based

Project overview:

Client is building a VR drone-simulation training tool on a MetaQuest 3 headset using bare-hand, image-based tracking (no controllers, no external sensors). A hand-based interaction prototype (anchor points on the hand mapped to keys) already exists. The team needs custom gesture recognition built for drone flight control (movement, roll, pitch, yaw) — this logic doesn't exist yet and needs to be designed from scratch.

Scope confirmed with client: this role covers both (1) true custom computer vision — working directly with raw image/camera data rather than relying solely on the headset's native hand-tracking output, and (2) custom gesture classification logic built on top of that tracking data. Both layers are in scope.

Core responsibilities:

• Build custom CV models/pipelines for hand tracking from raw camera input (beyond what the Quest's native SDK provides)
• Design and implement custom gesture classification for drone flight controls (movement, roll, pitch, yaw)
• Collaborate with the Unity/VR lead on gesture-to-control mapping — undefined scope, invented as the project progresses
• Optimize tracking accuracy and latency for real-time VR use
• Work within/extend the MetaQuest 3 pipeline where useful, while building custom CV where native tracking is insufficient

Required skills:

• Hands-on computer vision experience — model training/inference on raw image data, not just consuming a pre-built SDK's output
• Gesture/hand-pose classification experience
• Comfortable with real-time, low-latency constraints (VR use case)
• Experience with XR/AR hand-tracking pipelines (Meta XR SDK, MediaPipe, or similar) a strong plus, but must go beyond SDK-level work

Nice to have:

• Background in robotics, drone control systems, or flight simulation
• Unity/C# familiarity (to collaborate directly with the VR dev, not required to own it)