Autodesk Work

Twenty years at Autodesk (2006–2026) as a Principal QA Engineer and SDET, embedded in the core product teams for Autodesk 3ds Max, Autodesk Maya and research-driven AI tooling.

Release automation and CI

  • A daily beta release, hands off. Every night at midnight, a system I built by hand in Python and PowerShell took the 3ds Max daily build, wrote the release notes from Jira, virtualized the build with Turbo and published it to the beta users through Centercode's API, on Akamai and later Amazon S3. The full story.
  • From a beta user's report to a closed issue, automatically. Every defect and feature request a beta user logged went into Jira through the Jira API, was routed by feature to the QA engineer who owned it, and was closed on Centercode by a Jira trigger that sent the notes back to the beta user. The full story.
  • GPU testing. Built an 8-GPU test server on a Cubix GPU expansion backplane, with NVIDIA Tesla and Quadro cards, for render testing and validating GPUs.
  • HPC testing. Tested 3ds Max rendering on a 256-CPU high-performance computing machine.
  • Did all of the daily 3ds Max builds, and owned the 3ds Max beta release cycle and the Python repository for release automation: set it up, got it running and maintained it, and wrote a lot of its code alongside other developers.
  • Built automation and API testing tools in Python and PowerShell, and integrated them into Jenkins pipelines with Artifactory.
  • For over ten years, authored and maintained the Jenkinsfiles and the Python inside them, keeping the results honest by triaging failures rather than muting them.
  • Set up Xray test management in Jira for regression testing, used Zephyr's regression features as well, and owned the 3ds Max Animation Regression Suite; worked with Jira, Confluence and Workday through their APIs and webhooks.

Autodesk Maya MotionMaker AI

Autodesk recently launched Autodesk Maya MotionMaker AI, a deep-learning Autodesk Maya plugin that synthesizes character locomotion from path inputs or sparse keyframes. Developed by Autodesk Research, this tool leverages neural networks trained on extensive motion-capture datasets to output deterministic, curve-editable animation data natively within Autodesk Maya.

As Principal QA Engineer and one of two embedded within the core product team, I collaborated closely with UX design and graphics engineering to validate early-stage feature architecture. I conducted comprehensive stress testing, regression testing, and usability audits, on Windows and Linux, to identify technical bottlenecks, propose optimization refinements, and introduce workflow enhancements that guaranteed both system stability and low-latency user interaction.

I supported executive demo readiness for MotionMaker at TechX 2025, where the Autodesk CEO presented it, and the product earned the Customer Experience Award at TechX 2026. In October 2026, Fast Company named MotionMaker in Maya to its 2026 Next Big Things in Tech list, in the Media, Entertainment and Social category.

Read the full story at Autodesk Blog and explore the core software environment at Autodesk Maya Product Details.

Autodesk Maya ML Toolkit

The Autodesk Maya ML Toolkit accelerates character animation and rigging by using machine learning to approximate complex deformations — skin, muscle, cloth, and corrective shapes — on resource-heavy character rigs. The core feature, the Autodesk Maya ML Deformer, learns deformation patterns from high-fidelity reference rigs to enable faster iteration at higher complexities, without sacrificing the visual quality that production pipelines demand.

Partnered with the development team to test all of the toolkit, and Maya's AI features, across Windows and Linux, from first development through production release, focusing on AI training functionality and deformer validation. Documented comprehensive test results and collaborated closely on defect resolution to ensure the toolkit shipped at production-ready quality.

Autodesk Maya ML Deformer documentation · Autodesk Maya

The 3ds Max and Maya beta program

Centercode was the platform that ran beta programs for all of Autodesk, a company with 7.79 million subscriptions at the end of fiscal 2025, and I was on the small corporate team of system-wide Centercode administrators: whenever a division needed a beta site, information or a custom setup, I helped set it up, and I was the one they came to. I owned the entire beta process for Autodesk's Media & Entertainment division, and I was the Beta Program Manager for 3ds Max and Maya. I was a stakeholder in the program's customer experience, from the vision to the implementation; supported and trained many of the people inside Autodesk who worked with Centercode; and set up the hard technical parts, the Jira integration, for many of the beta sites.

A daily beta release, hands off

I designed and built, by hand in Python and PowerShell, the system that put a new 3ds Max build in front of the beta users every day. Every night at midnight it:

  • took the daily build;
  • pulled the Jira fixes since the previous build, and the new features, into release notes;
  • virtualized the build with Turbo (formerly Spoon);
  • published it, with all its files, to Centercode through Centercode's API, configuring the file posting and the release through the same API;
  • served the downloads from Autodesk's CDN on Akamai, and later from Amazon S3.

The whole process was hands off.

From a beta user's report to a closed issue, automatically

I built the API systems behind every defect and feature request the beta users logged. A report went into Jira through the Jira API and was assigned to the QA engineer who owned that feature; they reviewed it, and when they finished, a Jira trigger closed the issue on Centercode and sent the notes back to the beta user. The loop was fully automated. I also recreated customer-reported issues and worked the debug analysis through with the field and development teams.

The runbook

I wrote the complete runbook for running the beta program, on Autodesk's internal wiki: every process, from posting builds to administering the program, step by step with screenshots and covering every detail, so that anyone could fill in as needed with no prior knowledge of the process.

QA on 3ds Max and Maya

On Maya and 3ds Max I was a senior QA engineer under the QA manager, across continuous two-week sprint cycles: test plans, regression strategies and validation matrices, risk-based planning with manual regression passes where automation could not reach, and each release's test schedule and risks. I owned the defect lifecycle end to end.

I worked on nearly every feature in 3ds Max, very closely with each feature's developer, implementing and reviewing fixes and verifying and validating all of it. I built 3ds Max locally to test fixes and review code (C++, MFC, the 3ds Max SDK), and, on that experience, I was generally the one on the QA team the others came to with questions and problems.

  • Nitrous, the 3ds Max graphics system: initial testing.
  • Water simulation in 3ds Max (computational fluid dynamics): initial QA lead.
  • Python in 3ds Max: with Christopher Diggins, worked on the Python integration from prototypes to first customer download. He wrote the C++; I tested the integration, did the UX and the QA, created the shipping demos, and reviewed the docs and examples.
  • GPU testing: built an 8-GPU test server, on a Cubix GPU expansion backplane with NVIDIA Tesla and Quadro cards, for render testing and validating GPUs.
  • HPC testing: tested 3ds Max rendering on a 256-CPU high-performance computing machine.
  • IPv6: built the first IPv6 lab for the Media & Entertainment division's network rendering system.
  • Stadium airflow: built a prototype computational fluid dynamics system for measuring airflow inside stadiums for sporting events.

Earlier: contractor, R&D and QA

Before those twenty years, I worked at Autodesk as a contractor in R&D and QA:

  • Character Studio and its Physique system, the tools used for the famous Ally McBeal dancing baby.
  • R&D projects: Autodesk Plasma, Autodesk gmax, and the initial implementation of JSR 184, the Mobile 3D Graphics API for Java.
  • The initial testing and implementation of mental ray in 3ds Max.
  • The VRML import and export system.
  • 3ds VIZ, the edition of 3ds Max made for architects and designers.
  • The initial implementation of the TCB rotation controller in 3ds Max.