I taught myself to run AI that predicts freight lanes, tracks my books, and powers my MBA, and to run it privately, on my own machine, without handing my data to anyone. Every tool on this page came from a problem I hit running my own business.
Brownsville, TX · Bilingual (English / Spanish) · MBA in Business Analytics, December 2026 · Open to Data / BI Analyst roles.
I've been behind the wheel since 2007, when I earned my CDL Class A.
For years I ran under my own authority, and because I'd always loved Excel, I started building what I'd call dashboards today: simple sheets that told me exactly what each load would pay after fuel and costs. That instinct to measure everything never left. In 2023 I moved my freight over to Landstar to cut the paperwork and focus on the driving and the building.
Then AI showed up, and I leaned all the way in. I taught myself where to go, how to code, and how to actually ship things. My crude spreadsheets became professional tools that show me what works and what doesn't, and I expand them every single day. Every problem I solve opens the door to the next idea, and chasing that next idea is the part I genuinely enjoy.
That's who I am: an operator who builds. I've spent 20+ years leading teams (a 20-year Army career, a 40+ member platoon) and grew my freight business from $100K to $300K in revenue with data-driven decisions. Now, with an MBA in Business Analytics, I look at any business the way I learned to look at mine: find where the hours and dollars leak, measure it, and build the thing that fixes it.
When I stepped into trucking and decided to buy my own truck, I made myself one promise: never stop looking for ways to improve the business. Chasing those improvements is where I picked up the ideas behind the automations and AI technology I use, and some of what I created along the way is on this page, live and working.
That habit changed how I look at every business, including mine. I don't see a mystery. I see processes that can be measured, decisions that can be sharpened, and hours that software can give back.
Every business loses hours and dollars somewhere. I learned to spot it because it was my money.
Excel, SQL, Python, PostgreSQL. If it matters, it gets a number.
The proof is below: tools built for problems I lived.
Every card is labeled for what it actually is. Live systems run today on real data. Prototype and Proof of concept are working builds that are not in production use. Client delivery was built to a paying client's spec. Nothing here claims a result that did not happen.
A load-prediction pipeline that runs on my own machine. Landstar load-alert emails are parsed into a Google Sheet, exported to CSV, and loaded into a PostgreSQL triangulation engine that analyzes roughly 14,000 historical loads over a rolling 12-week window. It refreshes 6 to 8 times a day and feeds live predictions into Lane Planner, the phone app I use to decide where to point the truck.
Freight towns are grouped into market areas: a hub city plus every freight town within 75 road miles, so "Fort Worth area" includes Dallas, Weatherford, and Hurst. A lane is one area to another, summarized over the 12-week window. A loop is a chain of legs that starts where the truck is and ends back in Texas, with the empty miles home charged at cost so a loop that strands me far from home does not look better than it is.
Every loop gets a confidence tier based on how many of the last 12 weeks its weakest leg actually posted and how many agencies posted it. Runnable means the lane shows up most weeks with several agencies behind it. Call ahead first means it posts some weeks. Gamble means it is thin. The pay math is the same formula I use to check my own settlements, so the ranking is in real dollars, not a score.
It is a small warehouse: a fact table of loads, dimension tables for locations and brokers, geocoding for new towns, and a rebuild step that re-derives the market areas whenever new data lands. The sheet keeps about 30 days of history. The database keeps everything, and the ingest never deletes.
For an MBA project I analyzed roughly 6 million shipments from the U.S. Commodity Flow Survey across air, rail, truck, and parcel, then determined the smartest mode by distance and shipment value. The result is a single-file interactive dashboard with the crunched market data baked in.
I combined fish landings, harbor tide readings, and offshore buoy data, three public sources in three file formats, into one PostgreSQL database with 8 tables, 57,740 rows, and 34 rules enforced by the database itself. Then I answered five business questions in SQL and tested whether the catch moves with the ocean.
A revenue and expense tracking system I built for my own trucking books, with a Google Sheet, Apps Script, and an iPhone shortcut all wired together. I log a load or a fuel stop from my phone and the books update themselves.
One of my first dashboards. An early proof-of-concept quote tool built from real historical load and pricing data to help a broker gauge what a lane should pay: rate per mile, fuel surcharge, and margin, in one number.
A working prototype of a web tool for shippers. The dock manager types in the order, picks the trailer, and gets back the complete load plan in seconds: a to-scale trailer diagram, the exact loading sequence for the forklift crew, strap counts by row, and weight and height checks, all before the first pallet leaves the rack.
I was picking up a flatbed load of roofing materials and the dock crew worked out the arrangement on the fly: what rides at the nose, what stacks on what, where the loose rolls go. All of that knowledge lived in people's heads. They got it wrong, and I spent two extra hours fixing it before I could leave legal. Get it wrong and you find out late, at the scale, at the strap check, or when freight comes back damaged. As the driver, I'm the one who straps it and hauls it, so I know exactly what a good load looks like.
A packing engine that knows the freight (pallet sizes, weights, what can bear weight, what crushes) and the rules of the road (payload limits, 8'6" height, center of gravity, strap working-load limits). It packs the trailer the way a good dock crew would: crush-sensitive foam at the nose, pyramid stacks of loose rolls braced at the rear, and the front row strapped as the backstop that braces everything behind it. Custom items can be added on the fly, and more than 60 automated tests keep the engine honest.
Every reload is forklift time, dock time, and driver detention. Every overweight or over-height surprise is a fine or a late delivery. This tool moves those mistakes from the dock, where they're expensive, to the screen, where they're free. That is the pitch. It has not been proven with a paying shipper yet, and I say so.
A Python tool that turns a messy point-of-sale export into clean data and a ready-to-read sales report with a single command. It standardizes mismatched dates and prices, drops duplicates and broken rows, and accounts for every row so nothing disappears silently.
A lightweight Python tool that takes the raw, messy CSV a point-of-sale system exports and turns it into three usable files. It runs on the Python already built into your computer, so there is nothing to install and no monthly fee.
You run one line and point it at your export. It standardizes every date format, strips the dollar signs and commas off prices, tidies product names, removes duplicate orders, and pulls out broken rows (missing IDs, bad dates, zero prices). Then it hands you back a clean CSV, a plain-English report (revenue by product, by month, and top customers), and a machine-readable JSON.
What used to be an afternoon of copy-paste and find-and-replace becomes seconds. Every row is accounted for (kept plus skipped plus duplicates removed always equals rows read), and the report tells you exactly why anything was dropped, so you can trust the revenue figure. Build it once, reuse it forever.
When a broker books my truck, they send a rate confirmation, the sheet that says what the load pays and where it goes. Often no Bill of Lading comes with it. The Bill of Lading is the document that travels with the freight: what is on the truck, who shipped it, who receives it, and the receipt the consignee signs on delivery. Without one I cannot get loaded, and I cannot prove I delivered. I hand the rate confirmation to an AI agent, it pulls out the shipper, the receiver, the pickup and delivery windows, the order numbers, the commodity and the weight, and lays them onto the standard industry form. What comes back is a print-ready PDF, ready to sign.
A team of AI agents that power my MBA work. They hunt down credible articles and documents for my research, so my assignments are built on the most accurate, current data. They also turn my reading into MP3s I can listen to while I'm driving, and I ask questions about what I'm reading through a chatbot from the road.
Want to talk about a data role, or about anything I've built? I'm easy to reach.