Computer Vision Intern

Company Overview

Contineu is building the next generation of AI-enabled operating systems for construction sites. We leverage computer vision technology to extract critical information about progress, quality, safety, and operational processes from helmet-mounted 360-degree cameras.

Job Summary

We are seeking a Computer Vision Intern to join our team full-time at our Bangalore office. This is a paid six-month internship opportunity. While we prefer in-person collaboration, we are also open to remote work arrangements.

Responsibilities

• Implement and iterate on stereo matching, depth estimation, and multi-view reconstruction (MVS/SfM), including epipolar geometry, triangulation, and bundle adjustment.
• Design spatio-temporal smoothing/registration techniques for stable reconstructions under dynamic site conditions by fusing per-frame depth over time.
• Build pipelines for point-cloud denoising, segmentation, and alignment (using ICP/pose graph) via Open3D/PCL.
• Conduct camera calibration (intrinsics/extrinsics), rectification, and handle lens models typical of wide fields of view (FoV) and 360° captures.
• Curate datasets from on-site captures and develop quantitative metrics (accuracy, completeness, scale drift) and benchmarking protocols.
• Optimize models for real-time inference on-device, at the edge, or in the cloud; identify bottlenecks and apply quantization/pruning when valuable.
• Collaborate with backend teams to version data/artifacts, design storage solutions for large 3D assets, and integrate services into existing APIs.

Qualifications

• Solid understanding of 3D computer vision fundamentals: camera models, epipolar geometry, triangulation, and depth/disparity.
• Practical experience through coursework, projects, or internships focusing on stereo 3D reconstruction or multi-view geometry (SfM/MVS) and point-cloud workflows.
• Proficiency in Python with PyTorch (or TensorFlow) and OpenCV.
• Familiarity with Open3D/PCL for point-cloud operations.
• Strong problem-solving skills, debugging capabilities, and experience in handling large image/video datasets.
• Knowledge of version control systems, specifically Git, and basic experiment tracking.

Preferred Skills

• Exposure to COLMAP and PyTorch3D.
• Experience with pose estimation, SLAM, ICP/registration, and pose-graph optimization.
• Knowledge of CUDA basics for acceleration and profiling.

Additional Information

Job Type: Full-time paid internship

Stipend: 25,000/month (50,000 if you choose to work in-office)

Duration: 6 months

Work Arrangement: Primarily in-person preferred, remote options available