10+ years building ML data tooling, drone operations software, and full-stack products. I combine product judgment with hands-on ownership of architecture, implementation, and release.
Lead UI Developer · Jan 2016 – May 2018
Led UI engineering for Siri’s internal ML data platform across multiple locales. Built training-data and evaluation workflows, including complex ontology navigation and labeling, with ML engineers.
Owned UI architecture, code reviews, testing, and deployment; mentored contributors as development expanded across teams.
UX Developer · Jan 2019 – Mar 2020
Delivered airspace-management software used by the Australian government, replacing a JavaScript proof of concept with a production Angular application.
Built drone telemetry and fleet-management interfaces with flight operations and flight-test teams, translating safety-critical requirements into operator tools.
AI integration · Theory of Constraints
AI can help one team produce more while the organization delivers less. More proposals, code, or content can flood a bottleneck with review requests, rework, and coordination—consuming the capacity that determines end-to-end throughput.
I apply the Theory of Constraints to find where AI can relieve that pressure and increase what actually gets delivered.
Map workflows through interviews and shadowing. Find queues, decision deadlocks, and coordination overhead. Protect the capacity of the bottleneck.
Choose between existing tools and custom software. Test on real tasks, establish quality checks and human review, and measure cycle time, delivered value, and rework.
Teach engineering practices as more roles begin building software. Clarify ownership and changing responsibilities. Build the trust to leave spare capacity where it helps the system.
Founder · Mar 2020 – Present
I founded and built Theory A to make the relationship between business performance and market expectations visible.
I own the product from customer research through implementation and deployment: custom charts, statistical analysis jobs, database migrations, and data-provider integrations.
Explore Theory A
Agent workflows & memory
Hermes is a personal AI workspace for turning research and recurring tasks into reusable agent workflows. It carries context across sessions, records decisions, and packages useful methods as skills so agents can pick up work and improve over time.
On-device computer vision
Scores Go boards from images using local computer vision and WebAssembly. Processing stays in the browser.
Embeddings & visualization
Built an interactive 3D explorer of GloVe word embeddings, turning semantic relationships between words into a visual interface.
ML education
Built a visual explainer of MNIST neural-network training, making core ML concepts easier to inspect and communicate.
Model comparison
Compares how AI models draw the same prompt, with generation times alongside the results.
AI product discovery and prototyping. Internal tools and workflow integration. Technical mentorship and organizational AI literacy.
dj@theory-a.com