Deep Learning Engineer (Multimodal Vision Platform)

Matroid; Iowa; On-site; USD 120,000 - 180,000; Full time

Benefits offered by this job

Free lunch

Gym membership

Hardware budget

Equity

Flexible hours

Health insurance

Work from home opportunities

Job summary

Matroid is seeking a Deep Learning Engineer to build the models, capabilities, and systems at the heart of our multimodal vision platform. You will rapidly take ideas from research to production, applying the latest techniques to deliver new platform capabilities for visual inspection across customers.

You’ll research, design, train, and deploy models for detection, segmentation, anomaly detection, tracking, video understanding, and 3D reconstruction, and optimize deployment from cloud GPUs to

Qualifications

• 3+ years of industry or academic experience in deep learning, computer vision, and software engineering.
• Strong software engineering skills in Python and PyTorch.
• Experience building ML systems and capabilities, not just scripts or experiments.
• Experience productionizing models, including inference optimization (e.g. TensorRT, ONNX) and serving (e.g. Triton).
• Experience with 3D reconstruction and scene representation (e.g. COLMAP, VGGT, 3D Gaussian splatting, world models).
• Open-source contributions to ML tooling, personal projects, competitions, or hackathons.

Responsibilities

• Research: Critically analyze the latest research in computer vision and deep learning and decide what’s worth adapting for use in our platform.
• Build: Design, adapt, and train models across detection, segmentation, anomaly detection, tracking, video understanding, and 3D reconstruction tasks.
• Engineer: Turn new models and techniques into platform capabilities that customers can train and deploy on their own data.
• Deploy: Optimize architectures and deployment systems for fast, cost-efficient inference, from cloud GPUs to edge devices.
• Collaborate: Work with product, infrastructure, and field engineering teams to drive models from idea to customer impact.
• Develop: Build models and capabilities at the heart of a multimodal vision platform.

Skills

Python

PyTorch

Software engineering

Video understanding

3D reconstruction

Education

Master's in CS/SE/Math

Bachelor’s with equivalent experience

Tools

TensorRT

ONNX

Triton Inference Server

COLMAP

VGGT

Job description

• We are looking for a Deep Learning Engineer to build the models, capabilities, and systems at the heart of our multimodal vision platform. This is a broad and deeply technical role, with significant ownership and impact
• You’ll rapidly take ideas from research all the way to production: apply the latest techniques to build and deploy new platform capabilities, solving visual inspection for our customers. You’ll work in a fast-paced environment alongside experts who are passionate about their work
• Research: Critically analyze the latest research in computer vision and deep learning, and decide what’s worth adapting for use in our platform
• Build: Design, adapt, and train models across detection, segmentation, anomaly detection, tracking, video understanding, vision-language, and 3D reconstruction tasks
• Engineer: Turn new models and techniques into platform capabilities that customers can train and deploy on their own data and use cases
• Deploy: Optimize architectures & deployment systems for fast, cost-efficient inference, from cloud GPUs to small edge devices
• Collaborate: Work with product, infrastructure and field engineering teams to drive your models & capabilities from idea to customer impact

Benefits

• Free lunch, snacks, drinks, and caffeine every day
• Gym membership with yoga and pilates classes next door
• Budget for whatever hardware you want
• Competitive salary and equity
• Flexible hours
• Dental and vision insurance with 100% paid premiums
• Work from home opportunities
• Health insurance

Solid oral, written, presentation, collaboration, and interpersonal communication skills Strong software engineering skills in Python and PyTorch; you care about the quality of your code 3+ years of industry and/or academic experience in deep learning, computer vision, and software engineering Excitement about the full range of this work, whether it’s building novel video transformer architectures, getting realtime segmentation running on a tiny edge device, or making 3D reconstruction work reliably on customer data Master’s in Computer Science, Software Engineering, Mathematics, or a related field, or a Bachelor’s with equivalent experience Familiarity with modern architectures and techniques: the latest foundation models, how they work, what you can do with them Experience building ML systems & capabilities, not just scripts & one-off experiments Significant computer vision and deep learning research experience Prior work applying computer vision to real-world use cases Experience productionizing models, including inference optimization (e.g. TensorRT, ONNX) and serving (e.g. Triton Inference Server) Experience with 3D reconstruction and scene representation (e.g. COLMAP, VGGT, 3D Gaussian splatting, world models) Experience building or operating multi-tenant ML products Previous work at high-growth technology startups Open-source contributions to ML tooling, cool personal projects, competitions, or hackathons