What I Learned During My Engineering Internship at Guild

When I accepted my internship at Guild, I expected what most people imagine when they think about an AI startup: a fast-moving environment, lots of acronyms, and the occasional feeling that everyone else knew something I didn't. What I didn't expect was how quickly I'd move from observing to contributing. Within my first week, I was conducting competitive research to inform product positioning, building AI agents based on that research, and eventually shipping features that became part of the product.

It was equal parts exciting and intimidating, but it also reflected something I came to appreciate about Guild. If you're ready to jump in, people trust you to do meaningful work. That trust ended up being one of the biggest lessons of my internship.

Coming into the summer, I assumed the hardest part of building AI products was AI itself. I thought that once you had the right model or prompt, the hard work was mostly done. Working on Guild's Trust and Safety capabilities completely changed that perspective.

One of my projects focused on what happens after an AI agent is deployed. What should happen if an agent behaves unexpectedly? When should it be flagged, hidden, or disabled? How do you communicate those decisions clearly to the person who built it? Those questions weren't really about machine learning. They were about building systems that people could understand and trust, especially when something didn't go according to plan.

Addressing these concerns about AI safety and trust is crucial for real-world applications, and understanding this helped me see the importance of responsible AI development. That experience changed how I think about AI.

Experimentation is forgiving. Production is not. It's easy to build something that works in a demo. Building something that people rely on every day is a completely different challenge.

Production systems have to account for uncertainty, edge cases, and moments when AI doesn't have the right answer. They need to fail gracefully, explain themselves clearly, and make it easy for humans to step back in when needed. I realized that the model is only one part of the product. Everything around it, from governance and observability to thoughtful user experiences, is what ultimately determines whether people trust it enough to use it. This lesson applies to any early-career engineer working on reliable, user-centered AI systems.

That same lesson showed up in my own engineering work. Early on, I submitted my first major feature as a single pull request with more than 8,000 lines of code. My teammates, especially one of the cofounders, Vincent, helped me understand that great engineering isn't just about the final result but about making your thinking visible and understandable. Clear, maintainable code builds trust and confidence in your work, which is essential for long-term success.

One of my favorite moments came later in the internship when I noticed our UI didn't show users which tools and integrations an AI agent had access to. I started prototyping the frontend because something was missing. A few days later, I discovered Vincent had already begun building the backend for the same feature. Without planning it, our work fits together almost perfectly. It was a small feature, but it reminded me that good engineering often starts with curiosity. If something feels incomplete, it's probably worth exploring.

More than anything else, this internship changed the way I think about software engineering. In school, success is usually measured by whether your code works. At Guild, I learned that success is measured by whether other people can understand it, improve it, and ultimately trust it. That applies just as much to engineering as it does to AI itself. This shift in perspective helped me prepare for the realities of professional software development beyond academic projects.

I'm incredibly grateful to everyone who took the time to answer my questions, review my code, and invest in my growth over the summer. I came to Guild hoping to learn how AI products are built, and I am leaving with a much deeper appreciation for what it takes to build products that people can trust, and I can't think of a better lesson to carry into the next chapter of my career.

Bio
Rebecca Waterson is a software engineer and storyteller with a passion for building technology that's both technically ambitious and deeply human. She grew up between San Francisco and Lake Tahoe, where, at 13 years old, she made a deal with her mom: earn a Junior National cut in the 100-meter butterfly, and she'd get a St. Bernard. She got the dog. Rebecca later swam for Brown University's varsity team while earning degrees in Computer Science and Literary Arts, where she also served as a Residential Advisor and a Computer Science teaching assistant. She is now pursuing a Master's in Computer Graphics and Game Technology at the University of Pennsylvania and is a Widjaja Engineering Entrepreneurship Fellow.Outside of work, Rebecca enjoys hiking, backpacking, and hosting a weekly game night with friends.