Senior Computer Vision Engineer: Aerial Imagery & 3D AI

Duranta; Seattle (WA); On-site; USD 160,000 - 275,000; Full time

Benefits offered by this job

Equity

Job summary

Duranta is seeking a Senior Computer Vision Engineer to join our Seattle-based team. You will help build a state-of-the-art computer vision pipeline, and develop models for aerial and oblique imagery using Mask2Former, SAM, ViT, and Swin.

You will run large-scale experiments with PyTorch and contribute to ML infrastructure and deployment optimizations. You will collaborate with leading researchers and engineers, reading latest papers and translating ideas into production-ready systems.

Qualifications

• 5+ years experience in ML / deep learning, ideally in Computer Vision
• Strong coding ability in Python/PyTorch
• Trained segmentation/detection models on real-world data at scale
• Experience with generative modeling and large foundation models
• Worked in startups or fast-moving teams and can wear many hats

Responsibilities

• Build components of a state-of-the-art computer vision pipeline for our AI system
• Develop CV models (Mask2Former, SAM, ViT, Swin) for aerial/oblique imagery
• Run scale experiments with PyTorch and DL frameworks
• Develop and maintain clean, modular ML infrastructure
• Collaborate on model deployment and inference optimizations
• Read and reproduce ideas from top-tier papers (CVPR/NeurIPS)

Skills

Experience in ML/DL for CV

Startup experience

Fast-moving team experience

Papers-to-implementation mindset

Tools

Python

PyTorch

Job description

Duranta is seeking a Senior Computer Vision Engineer to join our Seattle-based team. You will help build a state-of-the-art computer vision pipeline, and develop models for aerial and oblique imagery using Mask2Former, SAM, ViT, and Swin.

You will run large-scale experiments with PyTorch and contribute to ML infrastructure and deployment optimizations. You will collaborate with leading researchers and engineers, reading latest papers and translating ideas into production-ready systems.