AI, Computer Vision Engineer

Computer Vision Engineer — Industrial AI Platform (Full-time) Doaz Inc. | Seoul, Korea About Doaz Doaz transforms industrial expertise into actionable AI. We build vertical AI solutions for engineering-intensive industries—construction, shipbuilding, heavy industries, and geotechnical engineering—where precision, regulatory compliance, and domain knowledge are non-negotiable. Our products are trusted by Korea’s leading enterprises, including

POSCO E&C, Samsung Heavy Industries, Doosan Enerbility, and KT Estate . From automating geotechnical analysis with

95%+ accuracy

to generating safety risk assessments in

under 10 seconds , we solve problems that generic AI cannot reliably handle. Founded: 2023 Team: 20+ engineers and domain experts Vision: Industrial Knowledge → Actionable AI The Role We’re looking for a

Computer Vision Engineer

to build the visual intelligence layer of our industrial AI platform. You will develop systems that understand engineering drawings, extract structured data from technical documents, and support automated verification across multiple industrial domains. This is not a demo role. Our CV models run in production and process high-volume engineering documents. In our world, a

1% accuracy improvement

can translate to massive reductions in engineering hours and downstream rework. What You’ll Build

1) Drawing Intelligence Object detection and instance segmentation for engineering drawings (architectural plans, structural drawings, P&IDs, ship block diagrams) Symbol recognition across industrial standards (KS/ISO/ASME, classification society rules) Extraction of relationships, cross-references, and revision changes across drawing sets 2) Document Understanding Layout analysis for technical specs, engineering calculations, and regulatory documents Table extraction from material schedules, BOMs, and equipment lists Multi-format processing: PDF, scanned images, and CAD exports (DXF/DWG) 3) Compliance & Verification Visual verification for compliance requirements (building codes, safety standards, maritime rules) Automated comparison between design drawings and as-built documentation Defect detection for construction/manufacturing quality inspection workflows Domain Applications (Examples)

Construction:

floor plan analysis, certification review support, finish schedule extraction Shipbuilding:

block drawing interpretation, piping diagram analysis, weld symbol recognition Heavy Industries:

equipment layout verification, safety zone compliance, P&ID digitization Geotechnical:

borehole log interpretation, geological profile visualization What We’re Looking For

Required 3+ years in computer vision / deep learning with production deployment experience MS or PhD in Computer Science, AI, or related field Strong PyTorch proficiency; hands-on with detection/segmentation (YOLO/DETR/Mask R-CNN, etc.) Strong Python engineering; comfortable with Git, Docker, and Linux environments Ownership mindset; able to drive projects from research to production Preferred Document AI experience: OCR, layout analysis, table extraction (LayoutLM/Donut/PaddleOCR, etc.) CAD/engineering drawing domain familiarity Experience in regulated/industrial environments (construction/manufacturing/maritime) Multimodal AI (Vision-Language Models) research or applied experience MLOps: serving, monitoring, retraining pipelines Publications or open-source contributions Tech Stack (Typical)

Modeling:

PyTorch, HuggingFace Transformers, Detectron2 Optimization:

ONNX Runtime, TensorRT CV:

OpenCV, Albumentations Document/OCR:

PaddleOCR, Tesseract, LayoutLMv3, Donut, DocTR, PyMuPDF Infra:

AWS (EC2/S3/Lambda/SageMaker), Docker, Kubernetes Data:

PostgreSQL, Elasticsearch, Vector DBs (Pinecone/Milvus) Collaboration:

GitHub, Notion, Slack Why Doaz

Solve real problems:

Production systems used by real engineering teams Deep technical work:

Accuracy is a core product metric, not an afterthought Domain expertise access:

Work with engineers who have decades of field experience Growth stage:

Shape platform direction as we scale in Korea and expand globally Compensation:

Competitive salary + equity (impact-driven) Interview Process

Application Review (Resume + Portfolio) Take-home Technical Assessment (approx. 3 hours) Technical Interview (review + deep dive, 60 min) Apply Email

doaz@doaz.ai with: Resume/CV Portfolio (GitHub, papers, or project documentation) A brief note on why industrial AI interests you We review applications weekly and respond to all candidates. Doaz Inc. — Building AI that understands how industries actually work.