About

I have done the work, not just advised on it.

Ray Technical Consulting is an independent practice built on eight years of end-to-end ownership of a live company data platform - the pipelines, the warehouse, the ERP, the reporting and the production issues that arrive on a Friday afternoon.

What that ownership involved

One business, one platform, every layer of it. That scope is the reason this practice can take a problem from requirements to production without handing it to anyone else.

  • ETL and ELT processes across multiple production servers, benchmarked and re-engineered for throughput, reliability and cost.
  • API ingestion from more than ten business systems - shopfronts, analytics, payment and affiliate platforms, fulfillment and ERP - into a single analytics environment.
  • Streaming log and event data with Kafka and Azure Event Hub for near-real-time visibility that batch loads could not provide.
  • Modeled, version-controlled and tested transformation layers with dbt, orchestrated and documented for the people who inherit them.
  • Cloud migration of Microsoft SQL Server from AWS EC2 to Azure, including consolidation onto managed services.
  • SQL Server performance work down to the trace level, with monitoring and tuning tools, across production, staging and development.
  • Backup and recovery ownership for every server, because a platform nobody backs up is a platform that ends the company.
  • BI requirements and delivery across the major tools, and production reporting owned through to the executive read-out.
  • An ERP migration, including the modules that create, read, update and delete sales orders, customers, products, payments and refunds.
  • A fulfillment and shipping migration to a third-party logistics model, which cut shipping times by more than half while holding total fulfillment cost to under a ten percent increase.
  • Analytics that moved money: market basket analysis behind a twenty percent lift in cross-sales, and a model that reduced direct mail cost by targeting the right recipients.
  • Python automation that removed manual data collection and repetitive order, shipping and reporting steps.
  • LLM agent workflows in production, running on self-managed infrastructure over the company's own data.

How I work

Engineering with an outcome attached

Every engagement names the business outcome before the work starts. Data work that cannot be measured is a hobby, and it does not survive a budget review.

Honest about AI

I have shipped LLM agents in production - which is exactly why I will tell you when your data cannot support one yet. The foundations come first, and that advice is free of any incentive to sell you a model.

Tool-neutral

No partner commissions, no reseller margin. Recommendations come from hands-on evaluation across the major warehouses, transformation tools and BI suites - the tools that fit, not the ones that pay.

Operational fluency

The numbers above come from order flow, fulfillment and finance systems. That means the conversation starts in operational terms rather than in database terms.

One accountable owner

Assessment, architecture, build, handoff and support run through the same person. Nothing is lost in a handoff because there is no handoff.

Documented and handed over

Runbooks, tests and monitoring are deliverables, not extras. You should be able to operate and extend what I build without keeping me on retainer forever.

Technical background

Depth where it matters, listed plainly:

  • SQL Server / T-SQL - tuning, execution plans, traces, indexing, backups and recovery
  • Azure - Data Factory, SQL Database, Event Hub, migration tooling
  • AWS - EC2, QuickSight, migration off hand-built instances
  • Streaming - Kafka and Azure Event Hub
  • Transformation and orchestration - dbt, Airflow, version-controlled and tested models
  • Python - Pandas, NumPy, scikit-learn, Mlxtend, Selenium, BeautifulSoup
  • Warehousing - Snowflake and cloud warehouse evaluation and delivery
  • BI - Looker, Tableau, Power BI, QuickSight, Metabase, Cli