Senior full-stack developer

Andy Campbell

I meet challenges where they are. Finding solutions lights me up, especially where software meets the physical world: 3D, geometry, manufacturing. My passion for solving problems doesn’t change even as the domain does.

Technologies

  • C#
  • Angular
  • Entity Framework Core
  • ASP.NET Core
  • three.js
  • WebGL
  • NgRx
  • SQL
  • RabbitMQ
  • Docker
  • SignalR
  • SSE
CAD automation 5
  • Autodesk Inventor (COM API + iLogic)
  • SolidWorks
  • AutoCAD
  • Fusion 360
  • Autodesk Platform Services (incl. Forge)
CNC programming & automation 4
  • Alphacam
  • RouterCIM
  • Mastercam
  • WoodWOP
3D spatial math 5
  • Vectors
  • Quaternions

Matrices

  • translation
  • shear
AI integrations 7
  • MCP
  • Agentic tool orchestration
  • Multi-agent

Multi-provider APIs

  • OpenAI
  • Anthropic
  • Gemini
Custom software & integrations 12
  • MES (homegrown)
  • CPQ (homegrown)

ERP

  • Sage
  • Epicor
  • SAP

PLM

  • Autodesk Vault
  • homegrown

CRM

  • Salesforce
  • homegrown
Infrastructure 5
  • IIS

Azure

  • Web Services
  • Azure DB
  • Blob Storage
DevOps 11
  • Git
  • SVN
  • Jira
  • Linear
  • ServiceNow

CI/CD

  • Azure DevOps
  • Azure Pipelines
  • TeamCity
  • Octopus Deploy
  • Bamboo
AI development tools 3
  • Cursor
  • Claude Code
  • Codex

Selected work

01 / 03

RapidQuotes manufacturability

RapidQuotes manufacturability

Instant design-for-manufacturing checks on customer-uploaded parts.

The Challenge

Customers upload a 3D model and expect a price in seconds, but not every part will print correctly. Thin walls can deform when printing, and while we can detect the wall thickness, it’s not always obvious if the placement will cause issues.

The Constraints

We offer a range of technologies and materials, each with different tolerances for wall thicknesses. The UX needs to feel snappy and responsive when the user configures the part. A round trip to the API to calculate thickness is too expensive. We may want to add other DFM checks in the future.

A 3D model of a moulded housing shaded green, with red and orange patches marking walls thinner than the printing process can reliably build. Beside it, automated design-for-manufacturability checks for part size, model integrity, and feature size, with the feature size check showing a warning.

The Solution

A C# pipeline takes the uploaded STL, runs a wall thickness analysis using the Polygonica C API, and writes a glTF with wall thickness baked into the mesh as per-vertex data. The custom three.js shader renders the heatmap straight from that. Green is in tolerance, red and orange are too thin, and the customer can accept the risk or send the part for review.

  • C#
  • ASP.NET Core
  • Polygonica
  • three.js
  • glTF
  • Angular

Side Quest Quant

Side project, live at sidequestquant.com

Platform for building algorithmic trading bots through natural language.

The Challenge

Create a platform where LLM code is generated on the fly and hot loaded to run safely in a loop indefinitely. I provide the structure and extensibility; users pair with their favorite LLMs to provide the strategies and rules.

The Constraints

The code is untrusted, so it can be broken or hostile. It has to compile and run at runtime. It needs to behave the same way whether it’s testing against billions of historical records or a live market data firehose. Several LLM providers are in the mix, and I own the deployment end to end.

A dark application interface. On the left, an agent conversation with the LLM provider set to Claude and a model selected, showing a dispatched job with an ID and running status. On the right, a column of completed backtest cards, each with a line chart of its output.

The Solution

The platform provides the interface contracts and the structure to run them. LLM generated code is compiled at runtime with Roslyn, then run inside a constrained sandbox. Agents across several providers help draft and revise them, and a custom MCP server exposes the same tools to Claude Desktop and Cursor. The whole loop runs from idea to code to backtest to live.

  • ASP.NET Core
  • Angular
  • SQL
  • Roslyn
  • Auth0
  • Docker
  • MCP

Ruler Engine

Configurator platform that turns engineered-to-order designs into instant quotes and shop-floor-ready models.

The Challenge

Build a tool for manufacturers of custom, engineered-to-order products to quickly generate accurate quotes that can be turned into mechanical models and manufacturing drawings.

The Constraints

Easy to use configurator with low level graphics that designers can use. Generate pricing instantly. Use customers engineering rules and design intent. Quick and easy way to turn those quotes into models and drawings for the shop floor.

A collage of three configurator interfaces built on the same platform: a playground layout tool showing a 3D play structure with a parts list and live price, a pressure vessel designer showing a metal tank model with ASME design requirement fields, and a modular rack configurator showing a wireframe of stacked storage bays with bay dimension fields.

The Solution

A platform that allows engineering teams or implementation consultants to easily define hierarchal assemblies through C#. API for this rules model that feeds an Angular front end with low detail graphics for initial design. One click outputs that generate 3D models and drawings ready for the shop floor.

  • C#
  • ASP.NET Core
  • Angular
  • Three.js
  • CAD APIs

The arc

Always looking to expand my skillsets. I follow a pattern of spotting the pain points, then finding or building the solution.

I started from installing kitchen cabinets, moved to producing them, then CNC programming for large shops. The first code anyone paid me for was VBA automating my own job: machining wood doors on a CNC router.

By 2011, while going to school on the side, I was writing .NET add-ins for Inventor and AutoCAD, used daily by a fifteen-person engineering department. I went back to manufacturing engineering where I learned how to build systems for different business practices, and which abstractions survive contact with a real shop vs falling apart the first time someone has to cut a part from them.

Six years of design-automation consulting produced a platform to quickly build web based configurators that generated instant quoting and drove CAD drawings to the shop floor. Now writing production software at Stratasys. No computer science degree. An associate’s in computer information systems, and the rest learned on the job, solving real problems, building real solutions.

What I build now

Creating production software for industrial 3D printing, at Stratasys since 2022.

Primarily .NET and Angular, on customer facing systems where what gets ordered turns into a physical part as it moves through our shop floor and MES system. The work typically comes down to the same shape: more throughput, less time spent waiting. The numbers below reflect the quoting portal redesign that I led.

Revenue
$2.56M $3.89M
+52%
Checkout
15s 6s
60% faster

Contact

Are you building software where the output leads to a physical object? Interested in using WASM to safely run external or LLM generated code on your app? Or just want to nerd out about custom shaders or toolpath creation? Reach out, I want to hear from you!