How My JGM Internship Showed Me That UI, AI, and Testing Are All Part of the Same Product

By Zahyra Otero

Going into my internship at JGM Innovation, I wanted hands-on experience in software development and AI. What I didn't expect was how much the work would show me that UI/UX, frontend development, prompt engineering, automation, and testing all come together in one product.

A large part of my internship was spent working on Eclipse, JGM Innovation's AI-powered web application. One of the first areas I focused on was its user interface. When I started, the interface was much more basic, so I had the opportunity to help redesign it into something that felt more polished and intuitive. I worked on both light and dark mode designs, reorganized the chat history so conversations could be separated based on when they were used, and worked on interactions such as editing messages and managing previous chats.

I really enjoyed this part of my internship because I've always liked the intersection between design and technology. It wasn't just about making Eclipse look better; I had to think about how someone would actually use it. Something as simple as where a button is placed or how previous conversations are organized can completely change how intuitive an application feels.

What I didn't expect was that redesigning the interface would only be the beginning of my work with Eclipse.

As the summer continued, I became much more involved with the AI behind the application. One of my biggest projects was working on Social Post, an Eclipse skill that generates social media ideas for platforms like LinkedIn, Instagram, Facebook, and TikTok. Instead of having the AI generate generic captions, we wanted the feature to understand JGM's existing social media style and create content that felt appropriate for each platform. This meant thinking about everything from caption structure and hashtags to emojis, visuals, and the goal of the post.

Social Post also taught me that an AI feature is rarely just a prompt. What started as a prompt-engineering project eventually expanded into a workflow for maintaining JGM's past-post dataset. I also worked on frontend improvements, including individual copy buttons that made it easier for users to take a generated caption and actually use it. Seeing all of these pieces connect was probably one of my favorite parts of the internship.

Another project that challenged me was Skill Builder, a tool that takes someone's rough feature idea and turns it into structured directions that an intern or developer can follow. As I worked on it, I started encountering more and more situations I hadn't originally considered. What if someone wants to build for an application other than Eclipse? What if they're starting a completely new application? What if their feature doesn't need AI at all? 

There were definitely moments where I would fix one edge case, test the feature again, and discover another one. While that could be frustrating, it also became one of the most valuable parts of the experience. I became much more comfortable debugging, testing different possibilities, and realizing that the first solution isn't always going to be the final one.

Toward the end of my internship, I also worked on Skill Test, which provides a more consistent way to test Eclipse's AI skills. This project made me think differently about AI development because a feature can technically "work" while still giving an incorrect or poorly formatted response. Building AI isn't only about getting a model to respond; it's also about making sure that response follows the right instructions and continues working as the application changes.

Looking back, one of my biggest takeaways from JGM is how connected different areas of technology actually are. Before this internship, I tended to think about UI/UX, software engineering, data, and AI as more separate paths. Working on Eclipse showed me that one feature can involve all of them. You have to understand what the user needs, determine how the AI or application should behave, build the logic behind it, think about the data it needs, test unexpected situations, and finally make sure the result is actually easy for someone to use.

I'm leaving JGM with stronger technical skills, but also much more confidence in my ability to take an idea, work through the problems that come with it, ask for feedback, and continue improving it. I'm grateful that I had the opportunity to explore so many different areas throughout one internship, and I'm excited to take what I've learned with me into whatever I build next.