“When AI creates 3D, our job is to make it visible and make it move.”

Q. What does a graphics engineer do?

We write code that renders 3D on screen. It covers a lot of ground: the visuals you see in games, CG in movies, and, in our case, working with geometry in CAD. It’s a technology with a surprisingly wide range of applications.

Q. Where can you see the graphics team’s work in TRINIX?

When you enter a prompt, a 3D shape appears in the middle of the TRINIX screen. That’s exactly the part our team owns. To break it down, there are three things: we provide the tools AI uses to create 3D shapes, we manage the resulting geometry data, and we display it on screen. If the AI team builds what makes the AI think, we build what makes the result of that thinking actually exist and move as a 3D shape. Since everything the AI produces is 3D, the AI team and our team work almost as one unit. We’re responsible for the entire path, from the moment a shape is generated to the moment it appears on the user’s screen.

Q. What are you working on these days?

I’m focused on advancing Text-to-CAD, a feature where AI generates 3D shapes from natural language. Creating precise, detailed objects is one of TRINIX’s strengths. To take that a step further, speed is the key. Cutting generation time while keeping accuracy is my main challenge right now.

Q. Was there an object that was especially hard to handle?

A hammer. People usually guess factory equipment or large machinery, but objects made only of straight lines are quick to build, no matter how big they are. The hard part is shapes with lots of curved surfaces. A hammer looks simple, but its head is full of complex curves and surfaces. With shapes like that, it’s hard for AI to understand exactly how the shape is structured and to decide which modeling methods to use, and in what order. Smoothly connected curved surfaces are especially tricky. Getting them right means choosing the right modeling method and combining several operations, which takes a long time. That’s why I deliberately pick things like hammers and the hole-filled parts you see on construction sites for testing. If you find a way to model the hard shapes more accurately and quickly, the easy ones follow naturally.

Q. What’s the most memorable thing you’ve accomplished here?

The assembly feature. Take a trash can whose lid opens when you step on the pedal. The feature captures the joint information of objects like that so you can move them within their real range of motion inside TRINIX. I owned that project and built it from start to finish. When I joined, TRINIX could only create shapes. Nothing moved. Now, if you create a laptop, the joints are set up so you can open and close it on screen. Press a lever, and the connected parts move with it.

Q. You took on a big project on your own as a junior?

I think that’s one of our company’s strengths. If there’s a project that needs doing and someone who wants to do it, they’ll hand it over regardless of tenure. They trust each team member. I learned so much researching and building it on my own, not just technical knowledge but also how to approach the work itself. What makes me proudest is that this feature is now in the product and actually running.

Q. What is it like working with the team?

Before joining, I mostly developed on my own. NdotLight is the first place I’ve worked with a team, and the team showed me how to work efficiently without stepping on each other’s work. It made me faster not only when collaborating but also when working alone. We also had a session every Thursday called “AX Recipe Share,” where the development and design teams shared how they each use AI. Sharing knowledge with each other taught me a lot of useful things and boosted my productivity. It feels like the way we work keeps getting updated.

Q. What do you do when you hit something you don’t know or get stuck?

When something isn’t working, I grab whatever I can, whether that’s books or AI tools, and try every approach I can think of. But some problems still won’t budge. The truly hard ones that even AI can’t solve. Those I bring to my team lead. Somehow, once they reach the lead, they get solved like magic.

Q. How did you come across NdotLight?

It’s really rare for companies to hire junior graphics engineers. I found fewer than ten openings in total. NdotLight was one of them, and the job description caught my eye. Experience with OpenGL, game engine experience as a plus: everything I loved and had been doing was written right there. CAD felt like an exciting space too. So I applied almost without thinking.

Q. Is there anything you’d like to do next?

Until now, it’s been a cycle of building and reviewing internally. Once the product launches, real users will try it and give feedback, and I’m really looking forward to that. I want to take that feedback and turn it into features. Development with real users.

Q. Finally, if you were recommending NdotLight to someone considering applying, what would you say?

Opportunity. Here, if you want it, the opportunity is there. If you want to see something you built make it into the product quickly, I don’t think there are many places where that chance comes as fast as it does here.