Job description

ABOUT BOON Boon is the professional AI platform built specifically for construction. Founded in the San Francisco Bay Area in 2023 by product and engineering leaders from Samsara, Apple, Google and DoorDash. Boon is backed by leading Silicon Valley venture capitalists. Our AI agents embed directly into existing workflows, from preconstruction estimating to bid management. They automate the repetitive tasks that drain time and margins while surfacing the insights leaders need to make faster and more confident decisions. The result is measurable impact. Teams move faster, bids are submitted sooner, win rates increase, and costs are reduced. Boon enables construction companies to build more, generate more revenue, and grow with confidence. ABOUT THE ROLE We are 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 inference pipeline. This is a 80/20 research-to-production role. You will propose new architectures, run the experiments that prove or disprove them, and ship the winning models to real customers. You will have autonomy over direction and experimental ideas, staying aligned with the team and the company's research focus. This is not a role for someone who wants to be told what to build. WHAT SUCCESS LOOKS LIKE Within your first 12-18 months, the successful candidate will: - Push our production model to ≥95% accuracy across multiple trades and scopes - Design a genre-defining, novel architecture for construction drawing understanding - Publish a paper on the work at a top venue (CVPR, ICCV, ECCV, NeurIPS, or ICLR). We're committed to publishing; we may selectively not release weights or code WHAT YOU WILL DO RESEARCH & ARCHITECTURE - Design and evaluate novel multi-stage vision architectures for construction drawing understanding — perception, text-object association, and relational reasoning across elements - Drive architecture decisions: backbones, decoders, fusion strategies, loss functions, training regimes - Run rigorous experiments with clean baselines, ablations, and held-out evaluation on real construction drawings - Own supervised training and self-supervised pretraining strategies - Pursue research directions that compound accuracy across trades and scopes PRODUCTION SHIPPING - Take models from experimental notebooks to the production inference pipeline - Work hands-on with PyTorch, YOLO, SAM, DINO, and other modern CV stacks - Collaborate with ML engineers on deployment, quantization, and serving - Debug real failures on real customer drawings and close the loop into the next training run CROSS-FUNCTIONAL WORK - Collaborate with the synthetic data, annotation, and infrastructure teams to make sure experiments have the data and compute they need - Partner with engineering leadership on the accuracy roadmap and strategic direction - Write clean internal research reports so the broader team can learn from your work - Present findings, trade-offs, and recommendations to engineering leadership DATA & EVALUATION - Help shape what data we acquire and annotate, based on what the model actually needs - Define evaluation datasets and metrics that track progress honestly — not Kaggle-style leaderboard chasing - Identify failure modes on real customer drawings and design experiments that address them YOU ARE A GREAT FIT IF - You have 7+ years of computer vision research experience, ideally with a track record of published papers, open-source work, or production CV models - You have deep hands-on experience with multi-modal/vision transformers — segmentation, detection, or joint text+vision tasks - You have worked with modern vision transformer architectures like SAM, DINO, or similar foundation vision models - You can move from a research idea to a trained model to a production-shipped system with minimal hand-holding - You think about experiments rigorously — clean baselines, meaningful ablations, honest evaluation on real data - You have a point of view on architecture decisions and can defend it with reasoning and experimental evidence - You thrive on autonomy and set your own direction while staying aligned with team goals - You communicate clearly in English (written and verbal) and can collaborate during California business hours REQUIREMENTS - 3-7+ years in computer vision research (industry research lab, applied science team, PhD research + industry, or equivalent) - Strong track record of published CV research OR trained production CV models that shipped at scale - Hands-on expertise in multi-modal dense prediction (segmentation, detection, or joint vision-language tasks) - Production experience with modern vision transformer backbones (SAM, DINOv2/v3, CLIP, SigLIP, or similar) - Strong PyTorch fluency and experience training large vision models - Ability to move models from research to production inference pipelines - Strong fundamentals in deep learning: optimization, loss design, regularization, self-supervised learning - Fluency in English (written and verbal) NICE TO HAVE - Experience with Graph Neural Networks or relational reasoning architectures (we do not expect this — most CV researchers do not — but it is a meaningful plus) - Experience with text spotting, OCR, or scene text detection integrated with vision models - Experience with LoRA, adapters, or parameter-efficient fine-tuning of large vision models - Experience with self-supervised pretraining (MAE, DINO, or similar) - Experience with engineering/technical drawings, document understanding, or layout analysis - Contributions to open-source CV research - Published papers at top venues (CVPR, ICCV, ECCV, NeurIPS, ICLR) WHAT WE OFFER - Direct ownership of research on a problem nobody in construction AI has solved — the first foundation model for construction drawings - Autonomy to propose, defend, and run your own experiments, shipping the winners to real customers - Publication-friendly — we support our team publishing their work at top venues - Work at the intersection of computer vision research and real-world industry impact, in a trillion-dollar market - Collaboration with a world-class team from top tech companies and research institutions - Meaningful equity in an early-stage, well-funded startup COMPENSATION At Boon, we want to attract and retain the best employees and compensate them in a way that appropriately and fairly values their individual contribution to the company. With that in mind, we carefully consider a number of factors to determine the appropriate starting pay for an employee, including their primary work location and an assessment of a candidate’s skills and experience, as well as market demands and internal parity. This estimate can vary based on the above mentioned factors, so the actual starting annual base salary may be above or below this range. A Boon employee may be eligible for additional forms of compensation, depending on their role, including sales incentives, discretionary bonuses, and/or equity in the company. BOON IS AN EQUAL OPPORTUNITY EMPLOYER As an equal-opportunity employer, Boon is committed to providing employment opportunities to all individuals. All applicants for positions at Boon will be treated without regard to race, color, ethnicity, religion, sex, gender, gender identity and expression, sexual orientation, national origin, disability, age, marital status, veteran status, pregnancy, or any other basis prohibited by applicable law.

Requirements

Must-haves

Within your first 12-18 months, the successful candidate will:
Push our production model to ≥95% accuracy across multiple trades and scopes
Design a genre-defining, novel architecture for construction drawing understanding
Publish a paper on the work at a top venue (CVPR, ICCV, ECCV, NeurIPS, or ICLR). We're committed to publishing; we may selectively not release weights or code
Design and evaluate novel multi-stage vision architectures for construction drawing understanding — perception, text-object association, and relational reasoning across elements
Drive architecture decisions: backbones, decoders, fusion strategies, loss functions, training regimes
Run rigorous experiments with clean baselines, ablations, and held-out evaluation on real construction drawings
Own supervised training and self-supervised pretraining strategies
Pursue research directions that compound accuracy across trades and scopes
Take models from experimental notebooks to the production inference pipeline
Work hands-on with PyTorch, YOLO, SAM, DINO, and other modern CV stacks
Collaborate with ML engineers on deployment, quantization, and serving
Debug real failures on real customer drawings and close the loop into the next training run
Collaborate with the synthetic data, annotation, and infrastructure teams to make sure experiments have the data and compute they need
Partner with engineering leadership on the accuracy roadmap and strategic direction
Write clean internal research reports so the broader team can learn from your work
Present findings, trade-offs, and recommendations to engineering leadership

Nice-to-haves

Experience with Graph Neural Networks or relational reasoning architectures (we do not expect this — most CV researchers do not — but it is a meaningful plus)
Experience with text spotting, OCR, or scene text detection integrated with vision models
Experience with LoRA, adapters, or parameter-efficient fine-tuning of large vision models
Experience with self-supervised pretraining (MAE, DINO, or similar)
Experience with engineering/technical drawings, document understanding, or layout analysis
Contributions to open-source CV research
Published papers at top venues (CVPR, ICCV, ECCV, NeurIPS, ICLR)

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

aiml

Platform: ashby
Effort: quick

Benefits

Benefits detected in the listing.

Vision