The journey, as a commit history

Newest first. Twenty years, one repository.

1. feat: run Kubernetes, DevOps and GPU workloadsHEAD

   Closer to the platform than I've ever been: clusters, delivery pipelines, and CUDA on GPUs. It's the same job as before, getting the right work to the right place at the right time, with a different kind of inventory.

2. feat: go back to school for CS, statistics and mathFall 2021

   In fall 2021 I set out on a path through computer science, statistics, and mathematics, studying part time while working: algorithms, operating systems, computer architecture, number theory, cryptography. Fall 2026 has me in DevOps and GPU hardware.

3. feat: add AI to the planning stack

   Moved from classic demand planning toward AI-augmented planning platforms, where models propose and people decide.

4. feat: lead the teams that build it

   Managed engineering and architecture teams, and mentored consultants toward sharper technical judgment and the confidence to lead client conversations.

5. feat: design enterprise and cloud architectures

   Took on architecture for the platforms large organizations depend on, with TOGAF and AWS in the toolkit, and learned to talk trade-offs with executives as easily as with engineers.

6. feat: deliver ERP, inventory and demand-planning systems

   At Amazon, o9 Solutions, and IBM, across Oracle EBS and Fusion Cloud and SAP ECC and S/4HANA.

7. feat: architect homegrown supply chain systems

   Alongside the ERP work, I architected homegrown supply chain systems in modern programming languages.

8. init: build supply chain systemsinitial commit

   It started in operational chaos, which turned out to be an excellent place to learn what systems are for.

Back in the classroom since fall 2021

Twenty-three courses, taken part time while working, across NC State, UNC Chapel Hill, Colorado State, and Georgia Tech: algorithms, operating systems, computer architecture, numerical analysis, statistics, and cryptography. This term it's DevOps and GPU software and hardware, and next spring is advanced microarchitecture, machine learning, and computational complexity.

See the full coursework timeline

Break the cluster. It puts itself back together.

A tiny Kubernetes scheduler running in your browser: 13 pods across two CPU nodes and two GPU nodes. Click a node to take it down and watch the control loop reconcile what's wanted with what's running. Then turn on the autoscaler and break the GPU nodes.

web and api pods run on CPU nodes. ml pods need a GPU, and GPU nodes are tainted so nothing else can land there. Click a down node to bring it back.

Ship a change. Then see how far a bad one travels.

A delivery pipeline you can break on purpose. Push a change through build, test, scan, staging, canary and production, then switch the canary off and run the sneaky one again.

Off the clock

Cryptography, theory, and high-performance computing are my rabbit holes. Here is the oldest cipher there is. Turn the dial until the message reads as English.