Research Scientist, Computer Vision & Production ML

Attentive.ai; San Francisco (CA); On-site; USD 200,000 - 275,000; Full time

Benefits offered by this job

Equity

Visa sponsorship

Job summary

Attentive.ai builds AI for construction and field services. As a Research Scientist, you'll advance computer vision, deep learning, and NLP powering automated takeoff and estimation.

You'll be an early member of the US research team with real influence over directions and ownership of research problems end to end, taking models to production. You will tackle object detection, segmentation, and multimodal extraction on construction drawings, design robust evaluation frameworks, and stay

Qualifications

• 2–8 years of applied ML/AI research with a focus on CV and DL.
• Deep expertise in object detection, segmentation, OCR, and layout analysis.
• Experience training and deploying neural networks at scale with a production model.
• Strong Python and PyTorch skills across modeling, training, and data pipelines.
• Practical GPU and distributed training experience.
• Ability to evaluate current research and apply it to real problems.

Responsibilities

• Own research problems end to end: framing, data, experiments, ablations, and the shipped model.
• Advance deep learning models for object detection, segmentation, and structured extraction on drawings.
• Build multimodal systems combining visual layout, geometry, and text.
• Design evaluation frameworks to continuously improve the model.
• Stay current with CV and ML research and apply it to construction problems.

Skills

Computer vision

Deep learning

Python

PyTorch

Multimodal / vision-language

OCR / layout analysis

Education

Master's or PhD in CS or related field

Tools

SVG CAD familiarity

Cloud platforms (GCP/AWS/Azure)

Job description

Attentive.ai builds AI for construction and field services. As a Research Scientist, you'll advance computer vision, deep learning, and NLP powering automated takeoff and estimation.

You'll be an early member of the US research team with real influence over directions and ownership of research problems end to end, taking models to production. You will tackle object detection, segmentation, and multimodal extraction on construction drawings, design robust evaluation frameworks, and stay