Senior CAD Automation & AI Engineer

Aieng is a fast-growing engineering startup, based in Naples (Italy) helping industrial companies turn AI and digital tools into concrete product development advantages. We work on complex, high-stakes projects where technical depth and strategic thinking go hand in hand.

The ideal candidate is a Technical Leader capable of bridging the gap between complex software development and industrial engineering.

THIS IS NOT A ROLE FOR CAD USERS

We need software engineers who have spent years building systems that operate on CAD files programmatically — who understand parametric modeling logic, DWG entity structures, and drawing derivation pipelines as engineers, not as operators.

If you know which ObjectARX class to subclass to intercept an attribute write, why a parametric regeneration breaks when you batch-modify features out of sequence, or how a 3D model property propagates into a 2D title block attribute — read on.

• 7–12 years of specific CAD/PLM software development experience
• Senior / Principal IC (not team lead, not junior, not generalist)
• ≥ 3 CAD automation tools shipped in production (not PoC)
• CAD internals depth required — not API surface knowledge

• OEM in-house CAD/PLM teams: Airbus, Leonardo, Thales Alenia, Stellantis, Naval Group
• CAD-adjacent ISVs: Physna, Trimble, nTopology, PDSVISION, TechSoft 3D, CAD SaaS startups

• Build batch pipelines to extract, validate, and inject metadata into DWG/DXF files (title blocks, revision tables, drawing numbers, materials) across datasets of thousands of drawings — with structured validation loops and zero tolerance for silent errors.
• Write code that parses engineering PDFs with no mapping table provided — identifying layout zones, locating title block structures, extracting tabular data through geometric analysis, OCR, and vision models, without human supervision.
• Program against 3D CAD APIs to navigate parametric feature trees, inspect model attributes, and apply targeted corrections: BOM discrepancies, geometry violations, attribute mismatches.
• Integrate LLM-based extraction pipelines for drawing content where no structured source exists — the system must infer from the artifact itself.
• Connect pipelines to PLM environments (Teamcenter, Windchill, ENOVIA) via API; surface results in IBM DOORS and JIRA.

• Parametric modeling logic: feature tree structure, parent-child constraints, regeneration order — you build automation that handles failures correctly, not one that ignores them.
• DWG/DXF entity model: blocks, ATTDEF/ATTRIB entities, xrefs, layers — you have built title block automation that handles template variance without hardcoded coordinates.
• Drawing derivation pipeline: you know how 3D model properties propagate into 2D drawing attributes and have automated this propagation correctly.
• Geometry kernel awareness: working knowledge of ACIS/ShapeManager, Parasolid, or CGM; B-Rep topology at the level required to write geometry validation code.
• Assembly and BOM logic: component instances, transformations, configuration-aware BOM extraction.
• Revision and lifecycle management: how maturity states and change records map to CAD metadata — and how to respect them in automation.

• Language: Python (production-grade) + C++ (ObjectARX / CAA) + C# or VB.NET (SolidWorks)
• 2D CAD API: ObjectARX, ezdxf, pyautocad, AutoLISP, Autodesk Forge/ACC
• 3D CAD API (≥1 in depth): CATIA V5 COM or CAA RADE / SolidWorks API / PTC Creo Toolkit / NX Open
• AI integration: LangChain / LangGraph, OpenAI / Anthropic API, PyTorch, pdfplumber

• E1 — A batch DWG/DXF automation tool deployed in production (not PoC) — scale, error handling strategy, validation approach.
• E2 — An ObjectARX application, CATIA COM integration, or equivalent that a real engineering team depended on — which API, what it automated, at what scale.
• E3 — A pipeline you designed to extract structured data from engineering PDFs — how you handled layout variance across drawing templates.
• E4 — A CAD-to-PLM integration via API (Teamcenter, Windchill, ENOVIA or equivalent) — what broke first and how you resolved it.
• E5 — A validation loop for automated CAD modifications — how you measured and tuned the false-positive rate.

Full remote or local (preference for Naples, Turin, Rome) candidates will be considered.

Fluent English is required as well as Italian.