CAD scripting Expert

Responsibilities: — Design expert-level Terminal-Bench tasks by converting real, paid mechanical CAD work from your own experience into challenges that an AI agent must solve completely. — Precisely define each task including files, systems, and environment to ensure AI agents can attempt tasks without ambiguity. — Develop working reference solutions using scripted CAD tools such as FreeCAD or CadQuery Python that correctly rebuild parametric parts after parameter changes. — Create automated checks and tests to objectively verify results, covering geometry, dimensions, tolerances, and file outputs like STEP or IGES. — Calibrate task difficulty so that an expert takes about 60-75 minutes to solve each task and explain how real-world mistakes would be detected by checks. — Iterate on tasks based on reviewer feedback, documenting design intent, tools, data, and rules for reproducibility and validation. Requirements: — Minimum 3 years of hands-on experience designing real mechanical parts that were manufactured or shipped. — Recent hands-on CAD experience within the last 3 years, with ability to specify systems, tools, data, and rules used. — Experience building or modifying designs through scripting (FreeCAD, CadQuery Python, OpenSCAD, Onshape FeatureScript), not just GUI. — Ability to read engineering drawings, tolerances, and GD&T, and explain detection of design mistakes. — Technical fluency with Python or another programming language and command line, plus familiarity with files such as CSV/JSON, APIs, or databases. — Availability to start around early November 2026 and work on a flexible, task-based schedule. Nice-to-have: — Experience with parametric design intent. — Contributions to open-source hardware projects. — Experience with Docker and Git. — Experience writing automated tests. — Experience with SolidWorks, Fusion, or Onshape APIs and STEP/IGES workflows. Benefits: — Task-based vetting process replacing traditional interviews. — Paid per task, with each task expected to take about 60-75 minutes. — Flexible, task-based work cadence with expected start around early November 2026.