Turn this role into an interview — a resume and cover letter built around what this employer wants.
Holliday Group, LLC in Sarasota, Florida seeks a Computer Vision Intern for ForgePlan, an in-house plan-review platform. This on-site/hybrid internship is paid and focuses on building the CV layer that reads flattened architectural PDFs, detects rooms and walls, and provides code-compliant overlay guidance.
You will own a component, defend your accuracy, and present results to licensed professionals. Candidates should have Python and CV library experience, a passion for messy data, and be
Holliday Group, LLC · Sarasota, Florida · Full-time internship (on-site / hybrid) · Paid
Construction drawings are still reviewed the way they were in 1985 — by a person, with a highlighter, page by page, hoping they catch the missing smoke detector before the inspector does. ForgePlan is our attempt to end that. We are building a system that reads a flattened architectural PDF the way an experienced plans examiner reads it — finding the rooms, the walls, the fixtures, the labels — and returns code-compliant MEP overlay guidance. The computer vision layer is the hardest part of that problem, and it is the part we are hiring for.
Holliday Group, LLC is a general contractor and construction manager based in Sarasota, Florida, working across commercial tenant improvement, restaurant, retail, warehouse, office, and luxury residential projects. ForgePlan is our in-house software effort: a plan-review assist platform covering four disciplines — Electrical, Life Safety, Mechanical, and Plumbing — against NEC 2023, FBC 2023 (8th Edition), NFPA 72, FMC 2023, and FPC 2023, with a Sarasota County jurisdictional focus. We are not a lab. We have real drawings from real jobs, real inspectors, and a real deadline, which means what you build gets used and gets judged.
Twelve months from now, the person in this seat will have taken at least one piece of the ForgePlan vision pipeline from "sort of works on a good scan" to "we trust it on a live job set." That might be symbol detection, wall and room extraction, text and dimension OCR, or the evaluation harness that tells us whether any of it is actually improving. You will not be shadowing someone. You will own a component, defend your accuracy numbers, and watch your output land in front of licensed professionals who will tell you exactly when it is wrong.
We would rather hire someone with three finished projects and honest error analysis than someone with a long list of frameworks.