Jonathan Stowell

Created a multi-dimensional data warehouse with conformed dimensions to support key business decisions including portfolio valuation, multi-channel attribution, and IRR. Implemented the model, ETL, monitoring, and logging from dev to production. Built in Snowflake, using Stitch and AWS Lambda functions.

In this system, I created a series of PL/SQL stored procedures and functions that managed the entire DW process, from fact and dimension creation to data cleansing and ETL. The project also included a bash shell framework to set up and manage the server components (i.e., FTP, etc.) of the data warehouse. Taken together, the system allowed for easily configuring a data warehouse system that was generalized but also customized for each client.

I was the principle architect. We first designed a metadata model and then reusable ETL components. Rather than a fixed set of data sources, this system had to support a fixed set of data source "type" as each client data feed could be different. The data warehouse model contained facts and dimensions that were conformed to meet the needs of all customers. This system, with is accompanying Tableau dashboards, was the principle means of measuring clients' digital marketing ROI.

While working on delivering services to a client, I saw the difficulty they had translating their business strategy into a data strategy. I envisioned and created a process and then a tool to make this possible to the non-technical business owner. I first created a low-cost prototype using advanced features of Google Sheets to demonstrate the value, then developed a metadata model and designed a user interface to make it more easily used and scalable. The basic idea is to capture the business logic in metadata and then generate application logic that implements this strategy in the application.

A Neo4j-based knowledge graph for a large pharmaceutical company that captures data to track the flow of materials from acquisition through manufacturing to distribution and many related functions.

It was transformational in its ability to provide never-before-seen insights into business processes. I was the primary architect of the data model and wrote an ETL framework that handles all transformations in a standard way that is easily maintained and easy to learn. I directed the creation of a GraphQL API on top of the graph, which is used by many applications. The database has over 2B nodes and consumes 2TB of data. It has delivered answers to never-before-answered questions and is becoming increasingly essential to the business.