Foundation Vision Scientist for Construction Drawings

Boon; United States; On-site; USD 180,000 - 260,000; Full time

Benefits offered by this job

Equity

Autonomy to publish research

Ownership of research projects

Job summary

Boon is building the first foundation model for construction drawings — a unified multi-modal vision system that reads, understands, and reasons about architectural, mechanical, electrical, plumbing, and structural plans the way a human estimator does.

As a Computer Vision Applied Research Scientist at Boon, you will own end-to-end experiments on our foundation model, from architecture design through self-supervised pretraining, supervised fine-tuning, and shipping production models into our

Qualifications

• 7+ years of computer vision research experience.
• Strong track record in multi-modal/vision transformers.
• Hands-on experience with PyTorch and modern CV stacks.
• Ability to move from research to production inference.
• Fluency in English (written and verbal).

Responsibilities

• Design and evaluate novel multi-stage vision architectures for construction drawing understanding.
• Drive architecture decisions: backbones, decoders, fusion strategies, loss functions.
• Run rigorous experiments with clean baselines and ablations.
• Own supervised training and self-supervised pretraining strategies.
• Take models from research notebooks to production inference pipelines.

Skills

CV research

Multi-modal vision

PyTorch

YOLO

SAM

DINO

Production deployment

Tools

PyTorch

YOLO

SAM

DINO

Job description

Boon is building the first foundation model for construction drawings — a unified multi-modal vision system that reads, understands, and reasons about architectural, mechanical, electrical, plumbing, and structural plans the way a human estimator does.

As a Computer Vision Applied Research Scientist at Boon, you will own end-to-end experiments on our foundation model, from architecture design through self-supervised pretraining, supervised fine-tuning, and shipping production models into our