Notes on building AI, data, and full-stack systems

Teague Porter is a software engineer based in Provo, Utah who builds AI-enabled workflows, production data platforms, and full-stack applications. This page collects working notes on how he approaches the craft — the patterns that keep showing up across shipping inference servers, dbt warehouses, and React apps that bill real money.

On AI and agents. The most useful AI features are rarely the flashiest — they are the ones that quietly remove hours of tedious work. Teague builds agentic workflows around real bottlenecks: an MCP server that lets a model reason over a 2,000-model dbt warehouse, or a personal training assistant that turns Garmin and 8Sleep data into recovery-aware recommendations. The pattern is consistent: give the model structured tools, keep a human in the loop, and measure the outcome in time saved rather than tokens spent. Fast inference matters too — a 3x throughput gain is the difference between a demo and a product people actually use.

On data platforms. A data platform is only as good as the trust teams place in it. At Qualtrics, Teague led a dbt medallion architecture across roughly 2,000 models, separating raw, cleaned, and curated layers so analysts could find a source of truth without a Slack hunt. The 70% processing speedups came from profiling first and rewriting hot paths second — not from rewriting everything. Good data engineering is mostly discipline: naming, layering, and making the pipeline boring on purpose.

On full-stack delivery. Teague enjoys owning ambiguous systems end to end. At Proxy he built a clean-energy platform as one of two engineers — backend, APIs, AWS, the React app, and the Stripe billing that turned it into $1M of annualized revenue. Shipping the whole thing taught him that architecture is a means, not an end; the goal is a product that works for the person using it. He reaches for Python, SQL, TypeScript, React, dbt, and whatever LLM gets the job done, and he is happiest when a Monday problem is gone by Friday.