About
A structural engineer for your data platform.
Jesse Smit
Founder, Smitster Data · LinkedIn ↗
I've spent 12+ years building, scaling, and rescuing data platforms — from SQL Server and Azure BI estates to enterprise-scale Snowflake and Databricks lakehouses, and the AI systems now being layered on top of them — for Fortune Global 500 and regional Fortune 500 clients across financial services and other large, regulated enterprises that can't afford to get this wrong. Smitster Data is the independent practice through which I bring that judgment to teams who need it.
My work sits at the seams: where ingestion meets transformation, where cost meets architecture, and increasingly where AI meets a data estate that was never designed to support it. I don't sell tools or resell cloud credits. I tell you what's actually going on, what it will cost you, and what to do about it — in language your engineers and your CFO can both act on.
Through incremental loading, infrastructure-as-code, and a handful of repeatable levers — Snowflake credit optimization, job-cluster tuning, CI efficiency improvements, consolidating legacy ingestion notebooks into clean, reusable frameworks — I have a consistent, repeatable track record of cutting cloud spend by a five-figure sum a month on enterprise platforms. I've also built AI-agentic tooling, via the Model Context Protocol and GitHub Copilot CLI, that reviews production code changes and helps teams ship faster without lowering the bar — grounded in documentation-first engineering, so the context an AI agent needs survives beyond any one person.
I keep client work confidential. You won't find logos or named references on this site, because the teams I work with value discretion — and because the quality of the thinking should speak for itself. The Resources page has write-ups on real, anonymized outcomes from recent engagements.
The stack I work in
Warehouses & lakehouses
Transformation & processing
Cloud & orchestration
BI & analytics
AI architecture
Tools change; the architectural judgment about how to fit them together is what carries over. That's what I'm hired for.
Track record
A decade-plus of platform work, in specifics.
Representative examples of the kind of work behind Smitster Data — real outcomes, real numbers, without naming the clients.
End-to-end architecture
I own systems from design through production, not just one layer.
I’ve built full BI platforms from the data warehouse up — source ingestion, ELT orchestration, semantic modeling, and reporting — deployed and maintained entirely through automated pipelines. My largest by table volume spanned over 4,000 dbt models, provisioned entirely as code with Terraform and Azure DevOps.
Performance turnarounds
Specializing in fixing what wasn’t working.
Rebuilt an underperforming Power BI semantic model from scratch in a week — replacing six months of a previous consultant’s work — and cut report render times from 2–3 minutes to under 2 seconds. Elsewhere, cut a 13-hour daily data-sync window down to 5 hours, and took ETL jobs that used to run for days down to minutes.
Cost & infrastructure optimization
Fast isn’t enough — it has to be cheap to run too.
Infrastructure optimization — VM and cluster right-sizing, credit usage, pipeline efficiency — is a consistent, repeatable lever: on one platform, it cuts recurring cloud spend by a five-figure sum every month. I’ve also right-sized a client off Azure Analysis Services’ most expensive tier, cutting that single line item by roughly $10K/month on its own, and built cost-visibility tooling elsewhere that let a client retire unused databases, reports, and pipelines outright.
AI & automation
Applying AI where it actually earns its keep.
Built and deployed an AI PR-review framework — powered by GPT-5-mini via MCP — that acts as one of two mandatory reviewers on every production code change, enabling autonomous, reasoned approvals without lowering the quality bar. Extended that into a multi-agent framework with a hierarchical skill library letting AI agents pick up and deliver backlog work on their own, and I’m now leading the ontology work to make the platform natively legible to AI agents, not just BI tools.
Get started
Ready to talk about your platform?
Send me a few lines about what's going on. If it's a fit, I'll reply personally with next steps.
Request a Consultation