About Me
I'm Curtis Todd, a simulation and pipeline engineer. For eighteen years I've built real-time systems and the tools that feed them, with a focus on procedural geometry.
I started my career co-founding and leading a game studio with friends. We became the only studio officially licensed by FIRST to build a FIRST Robotics game, which meant accurate robot simulation that still had to be easy to pick up and fun to play. It shipped on PC and Xbox 360 and reached over 250,000 players.
After that I spent time in healthcare technology leading an international team of engineers, QA, and product owners, then returned to real-time 3D, building physics-driven multiplayer systems and a reinforcement-learned agent hierarchy in Unreal, alongside procedural generation in Houdini and Substance.
Most recently, at Dream Book, I was the sole architect of a patent-pending pipeline that uses custom machine learning models to convert architectural floor plans into optimized 3D geometry for visualization and robotics training environments.
Now I'm looking for a full-time role building simulation environments, synthetic data, and the automated pipelines behind them.
Skills & Expertise
Machine Learning & Computer Vision
- PyTorch
- Computer Vision
- Deep Learning
- Reinforcement Learning
- Production ML Serving & MLOps
- YOLO / ResNet / SAM2 / U-Net
- Tesseract / EasyOCR / PaddleOCR
Engines & Programming
- C++
- Python
- Unreal Engine
- C#
- Unity
- .NET
- HLSL
Physics & Simulation
- Realtime Physics
- Bullet Physics
- Chaos Physics
- Multiplayer Systems
- Networking
Technical Art, Content & Design
- Houdini
- VEX
- Procedural Generation
- Performance Profiling
- Pipeline Optimization
- Shaders
- Substance 3D
- Maya
- 3D Modeling
Cloud, DevOps & Backend
- Docker
- FastAPI
- Celery
- PostgreSQL
- Redis
- CI/CD
- Git
- Microsoft Azure
- AWS S3
Leadership & Project Management
- Team Leadership
- Technical Mentorship
- Project Management
- Agile / Scrum
- Remote Team Management
- Stakeholder Management
Experience
Senior Content Pipeline Architect
Dream Book LLC
Architecting a cloud-native pipeline that turns 2D architectural plans and CAD models into game-ready 3D assets, combining computer-vision machine learning with deterministic procedural geometry, and authored the 79-claim patent behind it.
Key Responsibilities:
- Served as primary inventor on a provisional patent covering the computer vision and machine learning system converting 2D architectural drawings into structured 3D geometry.
- Architected a cloud-native, server-side 3D asset pipeline to automate the conversion of high-fidelity architectural models into optimized assets for real-time Unity applications using Python for orchestration to manage job stages and parallelize execution using Docker.
- Optimized the pipeline through inference batching, model-size reductions, and graceful degradation around unreliable stages, cutting processing from 5 minutes to 30 seconds and raising the success rate from 10% to 95%.
- Designed a headless procedural workflow using Houdini (SOPs/TOPs/VEX) to ingest and sanitize raw, unstructured geometry (DAE/FBX), drastically reducing the need for manual artist cleanup.
- Created a suite of custom Digital Content Creation (DCC) tools for Unity and Unreal that streamline environmental blockouts, reducing asset production time by ~85% and allowing the company to scale volume without increasing headcount.
- Closed the training loop in production by routing reviewed operator corrections into the training corpus, improving model accuracy through normal operational use rather than scheduled labeling drives.
- Authored task-specific validation suites measuring model output against downstream pipeline requirements rather than generic accuracy metrics, exposing regressions standard benchmarks concealed.
- Orchestrated parallel multi-stage inference on a two-core CPU with selective GPU offload, sustaining production throughput without dedicated inference hardware.
- Implemented automated performance protocols including intelligent geometry decimation, texture atlas generation, and draw call reduction to ensure assets meet strict mobile and web frame-rate targets while improving visual fidelity.
- Trained and evaluated the full computer vision model stack (YOLO, ResNet, Segment Anything, Tesseract OCR, EasyOCR) across the pipeline's detection, segmentation, and text extraction stages.
- Built an in-house annotation and training platform with a predict, correct, retrain loop where each annotation and training pass feeds the next through editable predictions compounding annotation throughput across projects.
- Handed off an authored master material with walkthrough sessions and ongoing support, establishing technical direction across team boundaries without formal reporting authority.
Senior Game Developer
Maverick Media Designs LLC
Built a deterministic, physics-driven multiplayer game in Unreal Engine, spanning hierarchical game AI, server-authoritative networking, and artist-facing content pipelines.
Key Responsibilities:
- Designed and trained a reinforcement learning motor-control policy for physics-driven robot agents in Unreal Engine, authoring reward-shaped training scenarios and integrating the learned policy for in-engine inference.
- Architected a hierarchical agent system layering a strategic director over the learned motor-control policy, coordinating a three-agent alliance through Unreal's pathfinding and decision-tree systems.
- Integrated the Bullet Physics C++ library into Unreal Engine through a custom native module, replacing the default solver to deliver deterministic, frame-locked simulation control.
- Engineered a physics-driven multiplayer solution with multithreaded client reconciliation, rewinding and resimulating the local physics sim within a 16.6ms frame budget while remaining server-authoritative.
- Built an Unreal Engine dedicated server that consumed less than 2ms of each server tick, hosting multiple matches per core with cross-platform play between Windows and Apple Silicon clients.
- Held an 8.3ms frame budget on a 2070 Super and 16.6ms on M1 Max by profiling CPU, GPU, memory, and VRAM, then trimming meshes, LODs, Nanite and Lumen, and material complexity.
- Modeled and textured a robotics part library in Maya to manufacturer specification and realtime performance budgets, serving a design tool that validates mechanical handoffs.
- Designed procedural tools for robotic assets in Houdini to accelerate artist-driven content creation, improve iteration speed, and maintain consistency across assets.
- Integrated skeletal targets, dynamic material instances, Vertex Animation Textures (VAT), and IK/FK control rigs to develop efficient, modular animation systems for mechanical assets.
Engineering Manager
ShareSafe Solutions LLC
Led international teams of engineers, QA, and designers delivering HIPAA-compliant software, modernizing development practices and cloud architecture while bridging business and engineering.
Key Responsibilities:
- Led and mentored diverse, international teams of engineers, QA testers, and designers to consistently deliver high-quality software updates and provide post-launch support for tens of thousands of users.
- Led the organization's migration to Azure DevOps, first automated build pipelines, automated testing, and cloud infrastructure without a failed transition or material disruption to delivery.
- Architected multi-tenant release management across isolated per-client environments, enabling selective beta rollout and independent client release schedules from a single production pipeline.
- Mentored engineers through the research, prototyping, and risk assessment stages of platform transitions, building project ownership capability alongside the technical outcome.
- Replaced a monolithic PHP/MySQL site with .NET microservices and migrated 20,000 user accounts across 9 clients from legacy self-hosted auth to OAuth 2.0.
Co-Founder & Lead Engineer
Diamond Bullet Studios LLC
Founded and led a studio that produced and published officially licensed FIRST Robotics video games, reaching over 250,000 players on PC and Xbox.
Key Responsibilities:
- Negotiated and managed an official licensing partnership delivering FIRST Robotics games to an audience of 250,000+ players on PC and Xbox.
- Collaborated with the FIRST Game Design Committee to create engineering-centric and interactive prototypes of confidential in-development robotics competitions.
- Built a custom game engine in C# with Microsoft XNA featuring a deterministic physics simulation for robotics gameplay on PC and Xbox 360.
- Created tools in 3ds Max using MAXScript to embed gameplay and physics data directly in .FBX assets, significantly accelerating game production.
- Cleared console certification with a single failure and hit an announced public release date with a two-person core team, meeting community expectation without slipping the window.
- Profiled CPU and memory on target console hardware with platform tooling, sustaining a locked 30 frames per second across gameplay and menus with no observed stutter.
Patents
Building Layout Extraction Pipeline with Rigid-Anchors and Notated Topology
2026Provisional Patent · Dreambook
A 79-claim system combining computer-vision machine learning with deterministic geometry generation to produce game-ready 3D assets from 2D architectural plans.
Education
B.S. in Computer Science
Southern Wesleyan University
Central, SC
B.S. in Management of Information Technology
Southern Wesleyan University
Central, SC