I'm a licensed electrical engineer building rule-driven automation for commercial design workflows. My background is in commercial building electrical design -- hotels, restaurants, offices, and mixed-use projects.
I am also building software to automate the repetitive, rule-following parts of that work -- grounded in code requirements, not just professional habit.
Ani-Revit2 is a C#/.NET 10 Revit add-in that automates circuit assignment and panel-driven load grouping inside Revit 2025/2026. It applies structured engineering rules during the design process -- not as a post-design check, but inside the live model where the work happens.
The core logic is Revit-free and testable. Circuit assignment strategies are explicit, configurable objects that can be inspected and traced back to the engineering requirements they implement.
A demo video is in progress showing rule-driven auto-circuiting and panel assignment running inside a live Revit model. The system is functional today -- a live walkthrough or deeper demo is available on request.
Building the Revit prototype clarified something important: encoding rule logic from professional experience alone was useful for prototyping, but not sufficient for a system that needs to be repeatable, auditable, and traceable back to code requirements.
ani-digital-twin-engine is that foundation -- a deterministic NEC/IBC calculation and knowledge engine, structured code library, and data ingest pipeline. Every calculation traces to a specific code section. No ML inference in the engineering path. The system includes a self-assessing validation layer that classifies its own data surfaces by trust status and explicitly labels what it has not verified.
Project / Building Inputs
|
v
Spatial + Program Understanding
|
v
Deterministic Rule + Calculation Layer
|
v
Revit-Side Workflow Automation
|
v
Governance / Standards / Reviewability
- Ani-Revit2 -- visible Revit-side workflow automation (the application layer)
- ani-digital-twin-engine -- deterministic, code-referenced logic and calculations (the foundation layer)
- ANI-Standards -- governance, conventions, and shared skills (the coordination layer)
- Supporting repos handle spatial reasoning, project-grounded test cases, and earlier calculation prototypes
Components exist at each layer. Some are mature and demonstrable; others are architecturally defined and partially built. The full pipeline is directional, not fully connected end-to-end.
That broader direction matters because the same approach can support practical, bounded automation inside engineering firms.
This work is oriented toward practical problems in engineering firms:
- Revit/API workflow assistants -- automating repetitive modeling, circuiting, and labeling tasks inside the design tool
- Electrical rule/calculation assistants -- bounded, code-referenced support for load calculations, branch circuit sizing, and panel scheduling
- Documentation/process assistants -- structured support for proposals, submittal prep, and repetitive engineering documentation
- Review/QA assistants -- helping organize and prioritize the review process without replacing professional judgment
The goal is not autonomous engineering. It is specialized, bounded workflow support for repetitive and reviewable tasks -- where the engineer stays in control and the system stays traceable.
AI agents assist in developing and maintaining this software, but they operate under explicit constraints -- and a human PE retains authority over all engineering decisions. The AI builds the tooling. Engineering judgment stays human.
What is different here is the combination: engineering domain expertise, deterministic rule logic traceable to building codes, and a governed AI-assisted development workflow -- applied to a real licensed-professional domain, not just a software demo.
- Licensed Professional Engineer (Electrical)
- Revit MEP, NEC analysis, electrical system design
- Building with C#, .NET, Python, and AI-assisted development workflows
- R&D informed by real project documentation and engineering complexity
Selected demos, walkthroughs, and architecture discussions are available on request. Repositories are private during active development.
Profile last updated: April 2026