Job Description

The Reality Labs Prototyping team builds the physical hardware behind Meta's XR and wearables research: CNC-machined, 3D-printed and molded parts, made in-house on fast turnarounds. We're looking for a computer vision intern to bring perception into the lab. That means catching defects before parts ship, confirming a part matches its CAD model, spotting trouble on a machine before it crashes, and giving the team a live view of the shop floor. You'll work with manufacturing engineers, machinists and software engineers, and your work will run on real machines and real parts.

Responsibilities

Build and train models that detect anomalies and defects in machined and printed parts from camera images.

• Develop methods that match captured images or 3D scans of physical parts to their source CAD models, to identify parts and compare what was built against what was designed.
• Prototype camera monitoring for CNC machines and 3D printers that catches setup errors, tool or fixture collisions and failing prints early enough to stop the machine.
• Deploy and maintain networked camera systems across the lab, covering capture, storage and the pipeline from camera to model.
• Build tools that generate and augment training data, including synthetic images rendered from CAD and automated capture rigs.
• Measure how the models perform on real lab data and share the results with the team.

Qualifications

Currently has, or is in the process of obtaining, a Bachelor's, Master's or PhD degree in Computer Science, Electrical Engineering, Mechanical Engineering, Mechatronics, Robotics or a related field

• Computer vision or deep learning experience from coursework, research or projects (detection, segmentation, anomaly detection, pose estimation)
• Experience programming in Python with a deep learning framework such as PyTorch
• Experience with image processing libraries such as OpenCV
• Intent to return to a degree program after the internship
• Must obtain work authorization in the country of employment at the time of hire, and maintain ongoing work authorization during employment 3D vision experience: point clouds, depth cameras, multi-view geometry, or aligning scans to CAD models
• Experience deploying cameras on a network (IP cameras, RTSP streams, edge devices such as NVIDIA Jetson) or running models in real time
• Experience generating synthetic training data (Blender, rendering from CAD, domain randomization)
• Experience with anomaly detection when there are few or no examples of defects
• Projects in industrial inspection, robotics or machine monitoring
• Comfortable working across disciplines, including hands-on with machinists on a shop floor
• Familiarity with CAD tools (Fusion 360, SolidWorks, NX) and 3D file formats such as STEP and STL