Eliuth Chavero Jasso

Forward Deployed AI Engineer | Engineering Systems Architect | AEC

I build production AI and software systems for complex engineering workflows — from customer discovery and architecture to deployment and measurable outcomes.

View Selected Work About Me Download Resume
LinkedIn · GitHub · Email

Applied AI  ·  Product Engineering  ·  AEC Technology  ·  Revit API  ·  Full-Stack Systems  ·  Electrical BIM/VDC

  1. Requirements
  2. Drawings
  3. Revit Model
  4. Evidence Graph
  5. AI Investigation
  6. Deterministic Verification
  7. Product Determination

Static input

Owner and code requirements, still in prose, still ambiguous.

Architecture / Systems

How I Think About Engineering Systems

Four layers, one continuous system. Click any node.

Sources

Intelligence

Proof

Product

Select a node above — see what it does and where it connects.

Reference diagrams

AI system architecture

User / Customer Problem

Domain Understanding

Structured Context

AI Investigation

Grounded Evidence

Deterministic / Governed Logic

Product Experience

Full-stack architecture

Revit / Desktop

C#/.NET

API

Python / FastAPI

PostgreSQL

React / TypeScript

Deployment architecture

Source

Tests

CI

Package

Installer

Production

Telemetry / Feedback

I don't just write code. I design systems.

Selected Work

01 — Engineering Intelligence Platform

Building a Governed AI System for Engineering Verification

ACT 01The Problem

Engineers manually review hundreds of project requirements across BIM models, drawings and project documentation.

Engineering evidence is fragmented. Decisions cannot be.

ACT 02The Investigation

Evidence is fragmented and decisions must remain auditable and reproducible. The system traces a chain — requirement, sheet, room, element, schedule, source — before it commits to an answer. The full evidence graph is explorable below.

Simplified pipeline

Revit + Drawings / PDFs + Owner Requirements

Evidence & Engineering Intelligence

AI Investigation

Governed Verification

Central Review Workspace

ACT 03The Proof

Population, coverage and contradictions determine whether a requirement is satisfied — not model confidence.

148Population (fixtures)
141Evidence coverage
3Contradictions
NOT METDetermination
∀x ∈ P,  h(x) ≥ 10ft Counterexample: element 114 — h = 8.5ft

Representative example — not client data.

ACT 04The Product

Everything lands in a working interface engineers actually use — production-ready, integrated directly into Revit.

TrackerEvidenceDeterminationReportHuman Review

My role

Technical ownership from product discovery through architecture, implementation, validation and deployment.

Stack

C#/.NET · Revit API · WPF/MVVM · Python/FastAPI · PostgreSQL · React/TypeScript · Applied AI · CI/CD

Engineering decisions

  • Deterministic decision authority
  • AI as investigation / reasoning layer
  • Evidence provenance
  • Multi-version Revit compatibility
  • Immutable evaluation runs
  • Human review separation
  • Local-first deployment
  • Release / installer validation

Evidence Graph

How a determination gets its receipts

Requirements, obligations, drawings, elements and findings as a connected graph — the chain a determination has to survive. Synthetic data, not a real project.

Drag nodes to rearrange. Hover to trace one hop. Click a Requirement to trace its full chain.

About

I'm a Forward Deployed AI Engineer and Engineering Systems Architect with a background in Electrical BIM/VDC and mission-critical infrastructure.

I spent several years working directly inside complex engineering workflows before moving into software and AI product engineering. That domain experience now informs how I design systems: I understand both the technical implementation and the people, documents, models and operational constraints behind it.

Today I build end-to-end products across desktop, backend, frontend, data, AI and deployment layers, with a focus on systems where reliability, evidence and traceability matter.

Based in Mexico · Working globally

Experience

Engineering → BIM/VDC → Technical Leadership → Product Engineering → AI Systems

  1. 2026 — Shokworks Forward Deployed / Product & AI Engineering
  2. 2025–2026 Crusoe Virtual Design Engineering / Critical Infrastructure
  3. 2024–2025 Salas O'Brien / Ehvert Hyperscale Data Centers / Electrical VDC
  4. 2022–2024 Comfort Systems USA BIM Coordination → BIM Lead
  5. 2020 — Electrical BIM Engineering foundation

Capabilities

What I Bring

Product & Forward-Deployed Engineering

  • Customer discovery
  • Technical scoping
  • Product definition
  • System architecture
  • 0→1 product development
  • Enterprise workflows
  • UAT / deployment
  • Technical stakeholder communication

Applied AI

  • Agentic systems
  • Local inference
  • Evidence grounding
  • RAG / retrieval
  • Evaluation
  • AI governance
  • Human-in-the-loop systems
  • Structured outputs

Software Engineering

  • C#
  • .NET
  • Python
  • FastAPI
  • PostgreSQL
  • REST APIs
  • React
  • TypeScript
  • CI/CD
  • Testing

AEC / Engineering Systems

  • Revit API
  • BIM/VDC
  • Electrical systems
  • MEP
  • Data centers
  • Engineering documentation
  • Drawings
  • Requirements
  • Digital twins

Writing

Notes from the field

Engineering AI Why LLMs Shouldn't Be the Final Authority in Engineering Verification Draft
Product What Building AI for Engineers Taught Me About Forward-Deployed Engineering Draft
Revit Designing Production Revit Add-ins Across Multiple Host Versions Draft
Agentic Systems From AI Assistant to Engineering Investigation Agent Draft
Evidence Building AI Systems That Can Show Their Work Draft
AEC Why BIM Data Is More Valuable Than It Looks for Applied AI Draft

Drafts in progress — links will go live as each piece is published.

Independent Research

SEION — spectral operators & structure-preserving mathematics

Exploratory, deliberately conservative research in operator theory, structure-preserving algebra and computational cosmology — published as reproducible code, papers and numerical certificates, separate from applied product work. Every claim is tracked with an explicit epistemic status, not asserted as proof.

A Spectral-Operatorial Formula for Polynomial Roots SEION · Operator Theory

A reproducible framework representing polynomial roots via a companion operator, an isospectral transport gauge, and numerical resolvent projectors on contours — framed as a local spectral projector bundle over coefficient space. Not a formula by radicals; not a contradiction of Abel–Ruffini.

github.com/voidzeit/seion-poly-roots →
SEION Math Core — Kernel-Integrated Laws SEION · Structure-Preserving Algebra

The mathematical nucleus of the Kernel-Integrated Laws framework: typed finite-dimensional n-ary laws, associator and symmetry defects, structure-preserving projectors, and finite cohomological checks. Claims are registered by status — definition, proved under assumptions, numerically verified, conjecture, open.

github.com/voidzeit/seion-math-core →
Constraints on an Ultra Late Transition in Dark Energy from Low Redshift Data Cosmology · Preprint

Analysis code and data products supporting a preprint constraining a late-time transition in dark energy using low-redshift observational data.

github.com/voidzeit/ultra-late-transition →

Selected Experiments

What I'm building next

Independent of client-confidential work. Source and demos go live as each one ships.

Revit Engineering Toolkit Open-source add-in utilities for Revit engineering workflows. In development
Engineering Requirement Compiler Generic implementation of a requirement parsing & compilation engine. In development
BIM Knowledge Graph Revit/IFC → graph visualization of model relationships. In development
Local Engineering AI A small agent that investigates a fictional engineering project. In development
Digital Twin Visualization A visual demo of live model & sensor data on a digital twin. In development

Interactive Demo

Ask an Engineering Project

Synthetic dataset. No client or project data is used.

Requirement “All exterior luminaires shall be mounted above 10 ft.”

Technical Philosophy

How I Build AI Systems

01

AI should investigate, not invent authority.

Use models for discovery, interpretation and reasoning; preserve governed logic for critical decisions.

02

Evidence before confidence.

Every meaningful conclusion should be traceable to the source that supports it.

03

Build for the real workflow.

Successful engineering software must fit the people, systems and constraints already operating in the field.

Resume

Resume — Product/AI Forward Deployed AI / Product Engineering Download → Resume — AEC Technology Engineering Software / AEC Systems Download →

Building something at the intersection of AI, engineering and the physical world? Let's talk.

Email · LinkedIn · GitHub

Available for selected product, AI and engineering-system opportunities.