I find the broken, expensive process and rebuild it with software, data, and AI.

Pricing six short-term rental units was manual and gut-driven. Set a flat monthly rate, bump it for holidays, hope for the best. It ignored booking pace, lead time, and demand, so some months filled at rates 15 to 20 percent below what the market would have paid, and soft dates were caught too late to fix.

A pricing engine that reprices the instant a booking or cancellation hits, through channel webhooks, plus a nightly sweep of the whole calendar. It reads each unit's two-plus years of history, live pace, lead-time curves, and local events, then picks one of four moves (observe, nudge, discount, raise) inside hard guardrails, so it never discounts a date that still books late. The edge is the decision logic, not the model: every date is priced against its own booking curve. It texts me the recommendation, I approve, and the new price pushes to Airbnb, Booking.com, VRBO, and the direct site in seconds.

And it learns. Every recommendation becomes a labeled outcome that feeds a Bayesian demand model, and the engine runs its own controlled price experiments (Thompson sampling) to learn how high it can go, not just how low. It generates its own training data and sharpens with every booking, a closed loop most enterprise tools never actually close.

Our best year ever was $233K in 2025. Halfway through 2026 we're already at $223K on the books, 96 percent of last year's full total, with our biggest revenue months still ahead. Roadhouse Lodge, the first property group I put on the AI pricing tool, is pacing 72 percent ahead of its historical average, and peak summer hasn't even hit. Across the full portfolio, every unit is pacing ahead of its historical average. Weekly pricing work went from a few hours to zero. It runs the calendar on its own, built and shipped to production by one person. What started on my own units is now in testing with a small group of other operators, and I'm validating a full year across new markets before opening it to the public. New properties come in through the free pricing assessments I offer hosts, which feed the engine real data as it grows.

Airbnb and VRBO take about 15 percent in combined fees, roughly $33K a year on my volume, for what is essentially a calendar and a payment form, and they own the guest relationship. Direct bookings were worse. We sent payment links after confirming, guests would not pay, reservations slipped through, and some guests showed up having never paid. Revenue just disappeared.

A full direct-booking platform with a pay-to-book model. The guest pays through Square at checkout before the booking even exists. The card is authorized, not captured, and the reservation goes pending. Only when I approve does the system capture payment, create the reservation in the channel manager so it blocks across every channel, send a branded confirmation with the right seasonal cancellation policy, and issue a self-service cancellation link. Decline, and the hold voids automatically and the card is never charged. Behind it, a webhook processor orchestrates the whole booking lifecycle and fires a pricing cascade so rates across all channels update within seconds of any booking or cancellation.

Fully live, with about $45K in direct revenue in 2025 and $40K already in 2026. 246 direct bookings at 100 percent payment compliance since launch, which made the unpaid-stay problem disappear entirely. Direct is now my second-largest channel, ahead of both Booking.com and VRBO, and it has saved roughly $13K in platform fees so far while giving me full ownership of guest data, messaging, and pricing. It also sets up the next move, hosting other operators' mountain stays on the same platform for a smaller cut than Airbnb.

The bank could not reliably scan its entire estate and identify which applications had Apache Struts embedded, the same vulnerability class behind the Equifax breach. Several teams had tried and failed. Without a trustworthy way to find every instance, the company could not report progress to its auditors or tell product owners what to fix.

I pulled my team together, reviewed exactly where every prior attempt had broken down, and mapped what we could do ourselves versus where we needed other groups. Then I reached across the org and worked directly with multiple scanning teams to define the criteria that determined whether Struts was actually embedded in an application. I built the SSIS and SQL pipelines that integrated the data from those security platforms into one reliable picture that finally met the auditors' governance bar.

A working enterprise-wide scan. Once it existed, the weekly report tracked the number of remediated Struts instances across the company, so leadership could finally show real progress and product owners got clear remediation instructions. The thing that had stalled for years got unstuck, mostly because I was willing to build the relationships across teams that no one else had.

I started this in college. I wanted an internship at a company that ran mobile user acquisition, studied their business model closely while interviewing them for school, and did not get the internship, so I built the same business myself and eventually partnered with them. We ran user acquisition for mobile games and apps across iOS and Android, including some of the biggest names in the App Store. Clients wanted burst campaigns that pushed an app into the top ten free, where organic downloads take over. Running those campaigns meant logging into advertiser systems all day, finding offers that matched our traffic by country, applying, launching, and babysitting them. It was eight hours of manual grind every single day.

A Python automation that hit the affiliate-network APIs directly, found the apps that fit our traffic by geo (Germany, Australia, the US, and more), and applied to them. Once approved, it loaded them into our system, set them live, and validated every offer, confirming each link was live, tracking conversions correctly, and integrated end to end before it ran traffic. It removed those eight hours of daily work outright. It worked well enough that I licensed it to other companies as a subscription product, productizing my own internal tooling. I also represented the company at industry summits in New York and Las Vegas each year, building the partnerships that fed the business.

Grew the company to $1.8M in revenue over four years with a team of four. Drove more than 1 million users across campaigns, including 500,000-plus to a single marquee title, and repeatedly hit the top-ten-free goal that unlocked organic growth for clients. This is the same pattern I run today with MileHigh AI, build automation to solve a real business problem, then turn it into something other operators pay to use.