R&D Engineer: 3D Vision & Reconstruction (Remote)

Inria; Valbonne; Hybrid; EUR 27,000 - 33,000; Full time

Benefits offered by this job

Public transport reimbursement

Leave: 7 weeks + RTT days

Teleworking possibility

Professional equipment

Cultural and sports events

Vocational training

Mutual insurance

Job summary

Inria Côte d'Azur University Center is seeking an R&D Engineer, focusing on computer vision and 3D reconstruction. You will develop methods to compare real-world 3D captures with references, aligning heterogeneous representations and detecting change robustly in varied capture conditions.

The role involves implementing 3D alignment, geometry-material-lighting decomposition, and training vision models on real and synthetic data, with integration into visualization tools.

Qualifications

• Engineering degree, Masters or PhD in Computer Science or related field.
• Experience training and evaluating deep learning models with PyTorch.
• Solid grounding in computer vision, including multiview geometry and modern vision models.
• Interest in joining an early-stage startup.

Responsibilities

• Design and implement 3D alignment methods combining visual/semantic correspondences with rendered geometry.
• Develop geometry-material-lighting decomposition for Gaussian Splatting representations.
• Detect and segment changes between captures or with digital models using images, depth, normals, and vision models.
• Train and evaluate models on real and synthetic data; integrate into visualization tools.

Skills

Deep learning

PyTorch

Computer vision

Multiview geometry

Startup interest

Education

Engineering degree or higher in CS

Tools

GLSL

CUDA

Job description

Inria Côte d'Azur University Center is seeking an R&D Engineer, focusing on computer vision and 3D reconstruction. You will develop methods to compare real-world 3D captures with references, aligning heterogeneous representations and detecting change robustly in varied capture conditions.

The role involves implementing 3D alignment, geometry-material-lighting decomposition, and training vision models on real and synthetic data, with integration into visualization tools.