Turning a Snowflake platform into attribution products, AI orchestration, and revenue at Community Impact. Previously $30M+ of operational impact at Applied Materials.
2015 first website → 2018 first work automation → 2021 founded a web studio → 2024 data scientist → 2025 data & AI lead
One unified system: designed and built that way from day one.
I'm a data and AI leader focused on collapsing the silos between infrastructure, applied AI, and revenue. I build end-to-end data ecosystems: from foundational warehousing through AI orchestration to the revenue products that use both.
At Community Impact, I lead the data and AI function for a hyper-local media company serving 50+ communities. In under a year, I've moved the organization from a zero-base to a production-grade data ecosystem: a Snowflake medallion warehouse built from scratch, an MCP-based AI orchestrator (Apollo) that lets Claude reason across the entire data layer, the company's first attributed-revenue product, and the technical backbone of its first signed data partnership. I also serve as CI's designated Texas data point of contact under the TDPSA.
Previously, four years at Applied Materials as a BI Analyst and Data Scientist, where I delivered $30M+ in operational and financial impact through ML models, Python automations, and Tableau dashboards across supply chain, finance, and engineering.
The throughline of my work: the platform, the AI layer, and the revenue motion are one unified system: designed and built that way from day one.
AI orchestrator for a data ecosystem
An MCP server letting Claude reason across a Snowflake warehouse, an ad-management system, and a CMS. Built on Anthropic's Model Context Protocol, designed so the data layer, the tool surface, and the model are all first-class. Extended with retrieval-augmented context over a document corpus.
Snowflake medallion warehouse, zero to production
Designed and built Community Impact's first data warehouse: a Bronze/Silver/Gold medallion architecture ingesting ad-management, ad-server, email, and CMS data into one unified, governed source of truth. The substrate every other CI data product runs on.
Do AI assistants cite your content?
A data product measuring whether ChatGPT, Claude, Perplexity, Gemini, and Copilot cite a publisher's sponsored content. Stood up ingestion of a new AI-crawler log source, lifted citation attribution from 21% to 91.4%, and shipped it as a customer-facing surface with a ~25x to 60x query speedup. A new analytics category for the business.
Finance-grade variance analysis on Snowflake
A company-wide executive dashboard that decomposes pricing and discount variance into reason-coded categories and reconciles charged revenue to finance, separating genuine discounting from earned terms and contractual comp. Built on a per-order-line Snowflake fact; surfaced and root-caused a seven-figure reporting discrepancy and traced it to source.
A production web app for advertisers
An authenticated portal where each advertiser sees their own attribution dashboard and monthly ROI report. I built it as its own app in the company monorepo, with tenant isolation enforced in Snowflake so one client can never see another's data, staff SSO plus magic-link login for customers, a review-and-send workflow for sales reps, and an operational kill switch. Built with AI coding agents under my direction.
First attributed-revenue product
A Streamlit application matching subscriber identity to ad exposures and transactions. Built at Community Impact and the technical foundation the company's first signed data partnership was built on.
Identity resolution at scale
CI's CDP layer with custom Snowflake UDFs (NAME_HASH, PHONE_HASH) consolidating subscriber identity across CRM, newsletter, and CMS systems, the substrate that makes attribution possible.
Self-service analytics for non-engineers
An internal portal that lets non-engineers (sales, operations, leadership) query governed metrics directly instead of waiting on a data team. Built in Next.js over a Snowflake Gold layer that serves as the single source of truth across every downstream app.
An agentic operating layer built with Claude Code
My own work runs through an operating layer I built with Claude Code on top of a markdown knowledge base. Nine scheduled agents write my morning briefing, rebuild the weekly roll-up, draft my Friday status email, turn meeting transcripts into project to-dos, reconcile my daily checklist against what actually shipped, and watch every repo for failed deploys. On its first run, the deploy watcher found repeated production deploy failures nobody had noticed. It's daily, hands-on proof that AI-native tooling can be built, trusted, and run in production.
$30M+ in operational impact over four years
Four years as BI Analyst and Data Scientist at Applied Materials. ML forecasting models, 20+ Tableau dashboards, and 10+ Python automations across supply chain, finance, and engineering. Cut manual review time by 80%+; recovered hundreds of operating hours per year.
Austin, TX
Senior Data Engineer
2025 – present
Austin, TX
BI Analyst · Data Scientist · Inventory Analyst
2021 – 2025
Austin, TX
Founder & Web Developer
2021 – present
Remote
SEO Account Manager
2021 – 2023
Austin, TX
General Manager · Assistant GM · HOA Manager
2018 – 2021
M.S. Analytics, Texas A&M · B.S., Texas State