This summer, Wet Dog Weather interns are tackling a unique challenge: building a weather application using AI and real weather data.
Antonio McElfresh, a meteorology student at the University of Oklahoma, chose to combine two of his biggest interests, aviation and artificial intelligence. His project, RuffRide, is an AI aviation weather app that helps pilots better understand turbulence. At the same time, a companion application aims to give airline passengers a simple answer to a frequently asked question: How turbulent will my flight be?
For Antonio, the project has been an opportunity to explore how AI can accelerate software development while learning what still requires human expertise.
Throughout the summer, he’s documented his progress in our AI Journal, sharing the challenges and lessons learned as he builds RuffRide with AI and Wet Dog Weather data.
From Storms to Aviation
Antonio’s interest in weather began with fear.
Growing up, strong storms were something he dreaded. Over time, that fear turned into curiosity, eventually leading him to study meteorology at the University of Oklahoma, where he’s entering his junior year while also pursuing a minor in Geographic Information Systems (GIS).
Along the way, he discovered another area of interest: artificial intelligence.
“I’m passionate about AI and how we can integrate it into our forecasting workflows and modeling to deliver better products,” Antonio said.
That combination of meteorology and artificial intelligence made the internship a natural fit.
Building an AI Aviation Weather App
Antonio’s primary project this summer is RuffRide, an AI aviation weather app focused on turbulence and flying conditions.
He’s also beginning work on a companion dashboard that simplifies turbulence information for airline passengers by answering one straightforward question: How strong will the turbulence be, and when will it happen during my flight?
Because aviation meteorology is one of his favorite specialties, the project allowed him to build something he would personally use as he explored new ways to communicate complex weather information.
Building RuffRide also changed how Antonio thinks about weather data.
“The weather data is generally easy to access,” Antonio explained. “It’s the method of displaying and communicating it that really matters.”
Learning to Build with AI
Antonio primarily used Claude Code throughout the internship. Having worked with it extensively before joining Wet Dog Weather, he already understood its strengths and limitations, allowing him to focus on building rather than learning a new tool.
For Antonio, the biggest advantage wasn’t simply having AI write code. One of his biggest takeaways has been that successful AI-assisted development depends on providing AI with enough context to understand the project before making changes.
Before implementing a new feature or dataset, he often asks Claude Code to “interrogate” the existing project so both he and the AI have a clear understanding of how the application is organized before moving forward.
That approach has allowed him to move from an idea to a working feature much faster than traditional development methods.
He also learned that using AI efficiently is a skill in itself. More capable models often spend additional time reasoning through problems, increasing both development time and token usage. Knowing when to provide detailed instructions, when to limit unnecessary work, and how to manage API costs became just as important as writing code itself.
AI Still Needs Human Guidance
Although AI has dramatically accelerated development, Antonio doesn’t believe it’s replacing software developers anytime soon.
Instead, the internship reinforced that human expertise is still essential.
He found that detailed prompts, careful planning, and ongoing review are necessary to produce polished applications. Small misunderstandings, unnecessary reasoning, or overlooked constraints can waste time and computing resources if left unchecked.
One example came while integrating the FlightAware API. By adding explicit instructions to avoid unnecessary API requests, he prevented the AI from making expensive calls during development.
Those experiences taught him that success isn’t simply about asking AI to build software. It’s about giving AI the right instructions, constraints, and context.
What Surprised Him Most About AI
Before joining Wet Dog Weather, Antonio had already experimented with AI-assisted coding.
This summer, however, he realized just how far the technology has advanced. Rather than manually setting up projects, organizing files, and configuring development environments, AI can now handle much of that process automatically. For RuffRide, he began with little more than an idea and a prompt.
He was equally surprised by AI’s ability to derive entirely new weather layers from existing datasets. During another project, AI successfully generated a sunset quality score using weather data, producing results that closely matched existing Terrier products.
At the same time, Antonio believes many people misunderstand AI-generated software.
He argues that the “AI slop” people often criticize isn’t an unavoidable consequence of using AI. Instead, it’s usually the result of relying on one-click website generators instead of thoughtfully building applications from the ground up. Used creatively, AI gives developers more freedom, not less.
Looking Ahead
The internship strengthened Antonio’s belief that AI will become an increasingly important part of meteorology.
Perhaps the biggest change, however, has been personal. The experience showed him that with today’s AI tools, a single idea can quickly become a functioning application. That realization has broadened his thinking about his future and the kinds of products he hopes to build throughout his career.
“I could be one big idea away from creating a product that changes the course of meteorology.”
As AI continues to evolve, Antonio is excited about the future of meteorology and the opportunities to build innovative tools, such as AI aviation weather apps, that help pilots make better-informed decisions.
This is the first post in our summer intern spotlight series. Check back next week to meet another Wet Dog Weather intern and learn about the AI-powered weather application they’re building.