Start with the problem
A sophisticated model without a useful business or human purpose is merely an expensive demonstration.
The journey
1978–1979
Data and early AI. My first IT work at the
University of Missouri's Urban Information Center involved
demographic and census data. In college, a group of us wrote a
checkers-playing program in LISP on a mainframe.
1980s onward
Forecasting and enterprise systems. At Southwestern
Bell and later AT&T-related roles, I worked with capacity
planning, mainframe utilization, predictive models, systems
management, and the practical realities of large organizations.
2010s–2020s
Cloud, data science, and machine learning. I
revisited advanced modeling as tools became more accessible, while
continuing work in technical architecture, cloud implementation, and
automation.
Today
Agents and multimodal AI. I work hands-on with
modern language models, retrieval and memory systems, image and
video generation, synthetic voice, music, and cross-platform
creative workflows.
What matters more than the buzzwords
Judgment beats mystique
You do not need to build the microwave to use it well. Deep theory helps, but problem framing, testing, and communication often decide whether a project succeeds.
Verify the result
AI systems can be impressive and wrong at the same time. Evidence, reproducibility, privacy, and honest limits belong in the workflow.
Current work
Agent systems: building and studying practical assistants with memory, retrieval, scheduled work, tool use, recovery, and human oversight.
Zara: a multimodal AI creator project that combines character consistency, voice, music, images, video, storytelling, and conversation.
Independent creative work: experimenting with original songs, visual narratives, and ethical ways for small creators to use advanced tools without pretending the tools are magic.
Recommended starting points
Think in decisions
Ask what decision the data or model should improve, and how you will know that it did.
Learn by building
A small end-to-end project teaches more than a folder full of untouched tutorials.
Keep the human in charge
Use automation to extend attention and capability—not to hide responsibility.
One book I still value is Data Science for Business, especially for connecting analytical work to business value. I also recommend the practical explanations of decision intelligence educator Cassie Kozyrkov.