Computer Vision Engineer for Symbol Detection on Construction Drawings

We are Volve, a company specializing in AI for construction, and we are seeking a Computer Vision expert to build a robust solution for symbol detection and counting on construction drawings and floor plans (PDF format). Our aim is to reliably find all occurrences of a template symbol (provided as PNGs) across multiple pages and return bounding boxes and counts with high accuracy. We have a collection of real drawings, template symbols, and some OpenCV-based prototype code, but need a more production-ready Python-only solution, with no limitation on open-source libraries. The ideal candidate would design and implement a symbol detection module that handles multi-page PDFs, varying resolutions, rotated/skewed drawings, and supports runtime template inputs without relying on thousands of labeled images. Approaches can range from advanced classical computer vision to hybrid or lightweight learned solutions, and we prioritize reliability and robustness above particular methods. Deliverables should include a Python module, documentation, and a short technical report. There is high likelihood of follow-up work for more symbol types and further integration if the initial prototype is successful. Please provide a brief introduction to your background, relevant project examples (preferably on technical drawings or CAD), and an overview of how you would approach the technical challenges outlined.

Budget range:
Large project ($3,000 – 5,000 USD)

About the client
AI for construction