Untangling enterprise chaos.Modernizing the cloud.Building the tools + teams that power it.
Hi, I’m Steven. I walk into tangled enterprise chaos, figure out how the pieces actually connect, and map what can be modernized — and exactly how it has to be executed. Then I help do the work: aligning leadership, getting every product team moving, and training the org so they can own and run it afterward. Here's some recent projects and things I've made for fun.
stack --core
- cloud
- Azure · Azure Local · AKS
- ops
- PowerShell · Terraform
- build
- TypeScript · React · Effect
- data
- SQL Server · Cosmos DB
- ai
- TanStack AI · OpenRouter
Untangle the mess. Align the org. Execute.
I never could pick a lane — I'll map a five-year modernization with leadership, then work shoulder-to-shoulder with the product teams who have to execute it.
See how it all connects
Map the dependencies and the history nobody documented — then figure out what can actually move, what should modernize, and the order it has to happen in.
Move the whole org
Get leadership and every product team pointed at the same outcome — and train people to run it long after the engagement ends.
Do the work, end to end
Migrations, architecture, automation, and full-stack builds — executed the way the system actually needs, not lift-and-shift.
Recent projects, plus a few I made for fun.
Azure Local Operations Platform
An evidence-first operations platform for 152 Azure Local clusters across five isolated environments. PowerShell collectors, WMI fault listeners, Windows Event Forwarding, and Azure APIs land in SQL Server 2025 — 130+ views, 230+ procedures — behind a typed React dashboard with a built-in AI assistant and governed automation.
TanStackEffectThe Forge
Real PowerShell from a browser — no shell exposed. Operators queue hash-pinned, allowlisted commands; a runner in each environment claims them from SQL, streams output to a live terminal, and leaves a log, result JSON, and ZIP for every executed run. The piece I'm proudest of.
EffectAudit trailAsk ALDHI
An AI assistant that investigates instead of guessing: 38 typed tools over 15 data sources, a sandboxed Code Mode, and answers that cite their evidence. No model-written SQL, no secrets in context, capped reads — and no email without a preview and an explicit send.
TanStack AIOpenRouterEffectEvent Intelligence
A rolling window of ~17M forwarded Windows events, turned into findings with researched meanings and frozen, replayable runs — without treating silence as health or proximity as cause.
Event DictionaryWEFAlphaForge — risk-first trading platform & AI Lab
A risk-first trading platform for crypto and ETFs. Strategies and an AI Lab only propose; a fail-closed risk engine decides; every fill lands in an append-only ledger. Paper trading on Azure; live trading stays off.
Read the breakdowndesign-doctor
A policy-driven design-quality scanner for React 19, shadcn/ui, and Tailwind v4 — 16 rules, a design policy the team owns, SARIF and GitHub output, and an agent skill. Pre-release; Tailwind language-service analysis is next.
Read the breakdown// More shipped work
Azure Local Redeployment Console
A scrappy PowerShell console — first version in about a day, hardened over the program — that kept the runbook in order and logged commands by site. One engineer, 96 Azure Local redeployments, two weeks.
Hyper-VFintech Analytics & Automation
Invoice PDFs in, reconciled numbers out: custom-trained OCR models in Python Azure Functions, T-SQL that links invoices and matches deposits and ACH payouts, and Next.js executive, cash, and risk views.
Document IntelligenceFiber Optic Project Management
Next.js 16 operations platform for fiber engineering — KMZ/KML route maps, segment workflows, POs, invoicing, payments, and payroll, with concurrency-safe money and Admin-only financials.
RBACVideo Frame Analysis Pipeline
Python GPU pipeline that scores every frame with smartphone-trained quality models, keeps every score in SQLite, and exports the best stills. Rebuilt after a benchmark caught the old models favoring blur.
pyiqaFFmpegThe bounty program.
Hundreds of legacy apps and a queue that would never clear — so I built a program that paid developers to modernize them on their own time. I pitched it to the CIO and HR, stood up the SharePoint tracker, and structured it so the company covered taxes and developers earned the full value of each migration.
It turned technical debt into a backlog people actually wanted — and it's the reason I got the only top performance rating IT had given in 15 years.
Got something gnarly to move, fix, or understand?
Azure migrations, monitoring you can trust, automation that should exist, or a full-stack build that needs an owner. Tell me what you're working through — I read everything that comes in.