Computer Vision, Applied Research Scientist
Boon Technologies, Inc. · Remote · United States (Remote) ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Fully remote (US hours)
- Computer Vision Research
- PyTorch + Transformers
- Publication support
- Senior/Lead level
You will design and evaluate novel multi-stage vision architectures to interpret architectural and engineering plans. The work involves moving models from experimental notebooks to production inference pipelines, ensuring they handle real-world construction data effectively.
Success means achieving high accuracy across multiple trades and potentially publishing findings at top venues like CVPR or NeurIPS. You will have autonomy over experimental direction while collaborating with engineering teams on deployment and data strategy.
What you'll do
- Design novel vision architectures for construction drawing understanding
- Run rigorous experiments with clean baselines and ablations
- Own supervised training and self-supervised pretraining strategies
- Transition models from research notebooks to production pipelines
- Collaborate with ML engineers on deployment and quantization
- Define evaluation datasets and metrics for honest progress tracking
- Present findings and recommendations to engineering leadership
What you'll need
- 3-7+ years in computer vision research or applied science
- Track record of published CV research or shipped production models
- Expertise in multi-modal dense prediction tasks
- Experience with modern vision transformer backbones like SAM or DINO
- Strong PyTorch fluency and large model training experience
- Ability to move models from research to production independently
- Fluency in English written and verbal communication
Nice to have
- Experience with Graph Neural Networks or relational reasoning
- Background in text spotting, OCR, or scene text detection
- Knowledge of LoRA, adapters, or parameter-efficient fine-tuning
- Experience with self-supervised pretraining methods like MAE
- Familiarity with engineering drawings or layout analysis
- Contributions to open-source CV research projects
Stack
- PyTorch
- YOLO
- SAM
- DINO
- Vision Transformers
How you'll work
- Fully remote position
- Collaborate during California business hours
- Autonomy over research direction and experiments
What you get
- Meaningful equity in an early-stage startup
- Support for publishing work at top venues
- Direct ownership of foundational AI research
- Collaboration with team from top tech companies