I spent 11 years keeping healthcare databases running at Centene and led 100+ SQL Server migrations with average downtime under an hour. Now I help teams make the same move to Azure, and build AI on top that checks its own work and tracks every dollar it spends.
SQL Server 2016 lost Microsoft support in July 2026. If you're still running it, you're either paying for Extended Security Updates or running without patches. Either way, there's a migration plan to write.
Start with an assessmentMost clients start with the assessment. You keep everything it produces, whether or not I do the migration.
Find out exactly what moving to Azure SQL, Managed Instance or Fabric would take, what it would cost each month, and where it could go wrong.
The move itself, done the way I did it 100+ times: rehearsed, measured, and reversible.
One real business workflow automated on Microsoft Foundry, with a second model reviewing the output and a log of what every call costs.
I built a platform on Azure that researches, writes, reviews and publishes daily and weekly market briefings on a schedule, with no one at the keyboard. AI is great at writing and bad at numbers, so every figure comes from a data feed, and a second model reviews every report.
| Daily Morning Market report | ~$1.15–1.85 |
| Weekly Week Ahead report | ~$5.90–6.94 |
| Writer input tokens per run, first build → now | 1.38M → ~0.3M |
| Runs needing at least one repair round | 16 of 16 |
| Projected cost of all three reports | ~$80 / month |
Estimated from per-call token logs and list prices; the Azure bill is the source of truth.
Built on Microsoft Foundry (Claude, GPT and Grok behind one vendor-neutral router), Azure Static Web Apps behind Cloudflare Access, Azure Communication Services email, Python and Jinja, run unattended on a schedule.
Every migration gets a dry run on a copy, a measured downtime window and a rollback I've actually tested. That's how 100+ moves averaged under an hour.
I build with AI coding agents under strict rules: one agent builds, a separate one reviews, tests must fail before the fix and pass after, and mutation testing checks that the tests catch real bugs.
Every model call is logged with tokens, latency and cost, so decisions like "which model writes which report" come from data, not guesses.
Built Python market-data systems and the Azure AI reporting platform above. Configured Azure SQL Database, Managed Instance and SQL Server 2022/2025 environments.
Led 100+ migrations across SQL Server 2000–2019, including multi-terabyte moves with average downtime under an hour. Architected Always On availability groups and failover clusters for mission-critical healthcare databases.
Kept servers, network and SQL Server at 99.9% uptime. Rolled out backup and DR processes that cut recovery time by 40%.
Off the clock: St. Louis Blues hockey and Classic World of Warcraft.
You get the person who has done 100+ migrations doing yours, not a junior team learning on it. For larger estates I scope the work in phases so nothing depends on one big bang.
Remote across the US, and on site around St. Louis and St. Charles County when it helps, for example on cutover night.
Yes. It tells you what you're running, what it costs to stay put, and what would break if you moved. Many teams use it to budget for next year.
Whichever fits the job and your rules. My own platform routes between Claude, GPT and Grok on Microsoft Foundry, and I can work inside your Azure tenant so data stays where your policies say it should.
A 30-minute call, no pitch deck. Bring your SQL Server versions, rough database sizes and whatever is keeping you up at night.