I've owned the full stack on enterprise-scale data — building the ETL, standing up governance, modeling the warehouse, and shipping the dashboards decisions get made on. Below, a live demo of AI-assisted querying — a pattern I've already run in production.
Built and ran ingestion and transformation pipelines across enterprise-scale record volumes — migrating legacy feeds to the cloud and leading an ERP data migration spanning 19 locations, 200+ tables, and 67 gateways. Consolidated redundant storage from 500 TB to 1.5 TB and kept daily loads reliable enough to report on.
Authored the data dictionary and governance documentation — standardizing metric definitions, naming, and lineage across analyst teams so a number meant the same thing everywhere it appeared.
Architected the data warehouse for self-service, then built star-schema semantic models with row-level security and optimized DAX. Moved models onto Fabric capacities and cut a daily refresh from three hours to 27 minutes.
Designed the KPIs and dashboards leadership actually used — an inventory-depth metric that lifted store availability 63%, and analytics work that drove a $100M inventory-cost reduction in the year after launch. Migrated reporting off Sisense to Power BI for ~$150K in annual savings, and set the design standards the whole org built against.
Built a natural-language-to-SQL layer with the Anthropic Claude API inside a production reporting tool — letting business users ask a question in plain English and get the query and the answer back. Proven effective in real use. The live demo at the top of this page runs the same pattern.
A white-label, multi-tenant SQL-passthrough reporting platform I designed and built end to end. An analyst certifies a query once against any SQL-accessible system of record; business users then run those reports live, on demand — no SQL, no IT ticket. The platform never stores customer data: reports execute against the customer's own system at run time.
I built a custom reporting platform in-house to kill the ticket queue for one operations team — then realized the same problem shows up wherever business users can't get to their own data. So I rebuilt it as a product.
An analyst certifies a query against the system of record. Business users run it on demand — always current, no SQL access, no ticket. The real competitor isn't Power BI or Excel; it's the IT request queue.
Parameterized SQL passthrough with strict per-tenant isolation — connection strings held per tenant in Key Vault under RBAC. Data Listo stores query definitions and run logs, never the customer's records.