Structured AI is building the AI workforce for construction design engineering.

Today, billions of dollars and months of human effort are wasted because engineers must manually check thousands of drawings for tiny mistakes that lead to massive rework. It's slow, painful, and deeply uncreative.

We're changing that.

Structured builds AI agents that understand complex mechanical, electrical, structural, and civil drawings — and automatically perform high-precision QA/QC, code compliance, and design consistency checks.

Engineers shouldn't be stuck hunting for missing tags at 2 AM. They should focus on designing the future.

Our long-term vision is bold:

become the go-to software platform for design engineers — from quality-controlling plans today, to generating and optimizing them tomorrow.

This means next-generation computer vision systems, autonomous workflows, and eventually true AI co-designers.

Structured AI | NYC | In-person | Can support OPT and CPT

The Role

You’ll turn research ideas into production-grade models that directly shape the product.

What You’ll Do

• Research and prototype ML approaches for automated clash detection in complex engineering drawings (MEP, structural, civil)
• Develop computer vision pipelines for detecting geometric conflicts across layered technical plans
• Design and evaluate models that understand spatial relationships, symbols, and annotations
• Work with large-scale PDF and BIM-derived datasets
• Build data curation, labeling, and evaluation frameworks
• Experiment with multimodal approaches (vision + structured metadata + rules)
• Optimize inference performance for real-world deployment
• Document findings and contribute to our internal research roadmap

You’ll work directly with founding engineers on problems at the intersection of:
computer vision, spatial reasoning, document intelligence, and real-world infrastructure systems.

What We’re Looking For

• Strong foundations in machine learning (deep learning, optimization, statistics)
• Experience with Python and ML frameworks (PyTorch preferred)
• Familiar with computer vision concepts (detection, segmentation, embeddings, etc.)
• Comfortable reading research papers and implementing ideas from scratch
• Curious, experimental, and willing to challenge assumptions

Excited about applying ML to physical-world systems (not just benchmarks)

We work across AI, computer vision, and full-stack engineering:

• SvelteKit + Tailwind
• FastAPI / Python backend
• Large-scale document processing
• Custom CV models (segmentation)
• OCR + multimodal LLMs
• Azure, Postgres, blob storage
• Distributed queues
• Revit, BIM360, Office Add-ins

If you like big PDFs, GPU workloads, clever agents, or turning messy drawings into structured meaning, this is your place.