Turn the data you already have into decisions that make an impact.
If your data is sitting unused, your analytics team is months behind, or the AI tools you bought went nowhere, that's the problem I take on.
I'm a senior data scientist and AI/ML expert you can bring in for one problem at a time. Pricing, forecasting, LLM pipelines, messy data, or the one big question nobody has time for. I find the real problem under the stated one. I don't guess which method fits: I run more than one, test the results, and keep the winner. Then I build the answer so your people actually use it.
Always fast. Always accurate. Always what you asked for. Always easy for you.
Real work. Real numbers.
Three things I built. Each one still runs today. Click one to see how it works.
Gone from one AI bill.
One pipeline had three hidden bugs. I found them and fixed them. Same output. Much smaller bill.
Rows, turned into prices.
One price per item, per location, from your own sales data. Each price says which rule is holding it back.
Reports. Zero by hand.
17 million survey rows turned into a report for every site. Checked against the old reports before anyone switched.
The old way is slow. And it costs more than it looks.
Your data team is months behind. A full-time hire takes half a year and a big salary. The AI tool you bought made slides, not decisions. Meanwhile, the question just sits there.
I work one problem at a time. We talk within a day. You get a plan within a week. You keep the plan, whatever you decide next.
What you get.
The method that wins, not the one that's trending.
Sometimes that's a price solver. Sometimes a forecast. Sometimes AI. Sometimes a two-page answer. I try more than one, test the results, and show you the reasoning. This one sets a price for every item and says which rule is holding it back.
A bill you can see coming.
I measure cost at every step before launch. The bill is a number on a screen, not a surprise. One pipeline I fixed was wasting $4,050 a month.
Fixes at the root.
I don't grab the biggest tool first. One staffing model was planning a line that never ends. The first fix was one formula. It shipped in days.
How I work. What makes it different.
I find the thing under the thing.
The retry loop under the big bill. The bad target under the long line. The real problem is rarely the one you were told.
I test methods against each other.
A wide toolbox, from constraint solvers to language models to reinforcement learning. I run more than one, measure the results, and keep the one that wins. No corners cut.
Built so people use it.
Maps, not tables. Models a team can follow. Proof the new thing matches the old thing.
Fast.
A reply in a day. A plan in a week. A working demo every week after that.
Runs where you need it.
Your systems or mine. Your team can run it, or I keep running it for you.
Talk to the specialist, directly.
No account managers. No junior handoff. The person on your calls is the expert doing the work.
How it works.
Three steps, and an optional fourth. You can stop after any one.
We talk.
We meet as soon as you're free. You tell me the problem, and we work out how you'll share the data.
You get a plan.
The real problem, the approaches I'll try, and the price. Yours to keep, free of charge.
I build it.
You see updates as soon as I have them. It runs where you need it, in your systems or mine.
I stay on.
I keep it running, tune it as your data changes, and take on the next question. A set number of days each month.
One senior AI and data science expert. On purpose.
I'm Jordan Storme. I design and run production data systems for an enterprise foodservice group, where every model has to hold up across hundreds of locations. I trained in cognitive science and computational modeling, graduated cum laude with the Dean's Award for Excellence, and did research in neurolinguistics before that. I'm based in Portland, Oregon, and work with teams across the US.
No account manager. No handoff. The person you talk to is the person who builds it.
More about me →Questions people ask first.
Do you work with our systems?
Almost always. I pick up new tools fast, so your stack sets the terms, not mine. Recent work includes:
- Languages
- Python · SQL · TypeScript · JavaScript
- Data
- Snowflake · Postgres · Supabase · dbt · spreadsheets
- Cloud
- AWS · Vercel · Cloudflare · Railway · Render
- ML and AI
- scikit-learn · PyMC · Snowflake Cortex · Claude and other LLM APIs · constraint solvers
- Reporting and apps
- Power BI · Streamlit · React · Shopify · monday.com
Our data is a mess. Is that a problem?
It's the normal case. The first look finds out what's usable now and what's worth cleaning. That goes in the plan.
Do we even need AI for this?
Maybe not. I'll say so. Language problems need LLMs. Pricing, forecasting, and scheduling need math. Math is cheaper, faster, and provable.
Who owns the code?
You do. It lives in your accounts from day one, unless you'd rather I host it.
How do you price?
Every plan comes with a fixed price. The build is fixed scope, so the price doesn't move. Ongoing help is a set monthly amount. You see every number before you commit.
Will you sign an NDA?
Yes. Most of my work is under one. That's why the case studies name no clients.
Let's talk data.
A few lines about your project is enough. I reply within 24 hours.
Let's talk data →