[ Reality → Model → Simulation ]
MEL TORRES// Operator-Engineer

Digital Twin & Simulation Engineer | Product Leader.

I turn ambiguous operational reality into runnable digital twins — and I own the roadmap, partners, and outcomes that make them get used.

Isaac Sim / OpenUSDROS 2 · Physical AIScan-to-BIMProduct · GTM · KPI
// LIVE: point-cloud → mesh → sim mesh
[01 // ABOUT]

One question, asked at progressively higher stakes.

My career has been one question asked at progressively higher stakes: how do you measure something real, model it, and make it better? I started in how we learn — curricula and learning analytics. I moved to how we work — human-performance analysis on Tesla's Fremont line, then AI-driven training systems. At CMU and HelioViews I moved to how we improve critical infrastructure itself — physics-informed digital twins of data centers and civil systems.

The method never changed: instrument reality, simulate it, optimize it with the best tools available. I build to prove, not to pitch — I'd rather ship a working closed-loop demo than a slide.

  1. 01 · LEARN
    How we learn
    Curricula and learning analytics.
  2. 02 · WORK
    How we work
    Human-performance analysis on Tesla's Fremont line, then AI-driven training systems.
  3. 03 · INFRASTRUCTURE
    How we improve infrastructure
    Physics-informed digital twins of data centers and civil systems at CMU and HelioViews.
[02 // PIPELINE]

From capture to simulation

  1. 01
    Reality Capture
    Lidar, imagery, site context
  2. 02
    Point Cloud
    Measured evidence
  3. 03
    BIM
    Structured model logic
  4. 04
    OpenUSD
    Interchange and orchestration
  5. 05
    Simulation
    Ops, robotics, digital twin
[03 // SELECTED WORK]

Selected projects

[01 // CIVIL BIM]

Culvert Scan-to-BIM

Reality-capture geometry translated into a clean structural model for inspection, coordination, and future simulation workflows.

  • Reality capture
  • Revit BIM
  • Alignment QA
Viewport Active
Interior reality capture brought into Revit
[02 // SPATIAL CAPTURE]

Indoor Reality Capture

Measured interior space reconstructed as a navigable twin, emphasizing geometry clarity and downstream utility over visual ornament.

  • Interior scan
  • Point cloud
  • Spatial context
Reconstructed interior brought into Revit with linked point cloud
Revit reconstruction
Cleaned point-cloud shell of the captured interior
Point-cloud shell
Elevation-banded point cloud showing vertical structure
Elevation map
Viewport Standby
HelioViews solar analytics report
[03 // SOLAR ANALYSIS]

HelioViews City

Site-scale solar analysis with building-specific orientation and sun-hour studies staged as a lightweight urban digital twin.

  • Urban model
  • Shadow study
  • Solar scoring
Viewport Standby
[04 // FACILITY OPS]

ThermalSight Drone Ops

A mission-control view of aerial inspection, thermal anomaly detection, and simulation-informed facility operations.

  • Drone telemetry
  • Thermal scan
  • Simulation
Thermal pass
Sim environment
ROS2 autonomy
Viewport Standby
[05 // REAL-TO-SIM]

Real-to-Sim Splat

A point-based reconstruction treatment that moves from diffuse capture to crisp, simulation-ready spatial intelligence.

  • Point rendering
  • Capture cleanup
  • Viewport studies
Viewport Standby
[06 // FLOOD TWIN]

CivicSense Flood Twin

Flood behavior visualized against terrain and infrastructure to support resilient planning, communication, and digital-twin storytelling.

  • Flood bands
  • Terrain context
  • Scenario view
Viewport Standby
[04 // CAPABILITIES]

What I build — and what I own.

ENGINEER ▸ I BUILD THE SYSTEM
Digital Twin & Simulation
  • Closed-loop OODA digital twins (Isaac Sim / Omniverse)
  • OpenUSD · USDPhysics · PhysX scene engineering
  • Discrete-event simulation + OptQuest optimization (FlexSim)
  • Headless validation at scale
Physical AI & Autonomy
  • ROS 2 closed-loop control systems
  • 6-DOF flight control (PID / predictive MPC)
  • Anomaly detection · fault-tolerant replanning
  • Sensor fusion: LiDAR / thermal / RGB-D / IMU
3D, BIM & Reality Capture
  • Scan-to-BIM, indoor + outdoor
  • Point-cloud-to-BIM authoring (Revit / Blender)
  • Gaussian splatting · photogrammetry · DEM
  • BIM→USD pipelines, cleanup & QA
PRODUCT LEADER ▸ I OWN THE OUTCOME
Product Strategy & GTM
  • Roadmap & product-line ownership
  • Customer discovery, GTM & launch
  • KPI & metric design
  • 0→1 prototyping to hardened production
Cross-Functional Leadership
  • Lead content, engineering & analytics teams
  • Partner integration & vendor relationships
  • Fluent with operators, engineers & executives
  • Pre-seed positioning (NVIDIA Inception)
Decision-Grade Delivery
  • Metrics leadership acts on (JPH · PPM · error/drift)
  • Real-to-sim data sourced to spec
  • Interpretable, human-in-the-loop by design
  • Build to prove, not to pitch
[05 // PRODUCT LEADERSHIP]

Ownership across the product line.

Roadmap & strategyproduct-line revenue & satisfaction ownershipcustomer discoveryGTM & launchKPI / metric designcross-functional team leadership
HelioViews
Founder & CEO

Own product end-to-end — roadmap, customer/GTM targeting, partner integration, and a 3-year scaling roadmap — while driving pre-seed positioning on the NVIDIA Inception track. Delivered a working city-scale MVP within a 30-day sprint.

XR Professional Development
Chief Product Officer (3.5 yrs)

Directed product strategy for an enterprise XR training platform; owned the roadmap and led cross-functional teams across content, engineering, and analytics; integrated NVIDIA enterprise AI tooling (ACE, NIM, Tokkio, Audio2Face) into shipped products.

LinkedIn Learning
Product Owner, Data Science Learning Library

Owned a product line for launch — content roadmap, learner satisfaction, and engagement — driving an 85% improvement in satisfaction and completion.

Tesla
People Analytics & Performance Consultant

Built performance- and retention-analytics frameworks adopted into Tesla's Employee Performance Program.

Cross-functional fluency · Speaks to operators, controls engineers, and executives in their own terms — most of what makes a simulation actually get adopted.

[06 // CONTACT]

Let's build something measurable.

Need someone to build the simulation, or to own the product around it? I do both — and I'd rather show you a working closed-loop demo than a deck. Zero-friction options below.

Or copy →
[ CREDENTIALS // FIELD STRIP ]
M.S., AI Engineering & Digital Twins — Carnegie Mellon University (May 2026)·FAA Part 107·NVIDIA Inception track·U.S. Citizen