Designing the Future with AI Simulation

What we talked about.

In this episode of the Sunny Ray Show, host Sunny Ray talks with Anika Stein, CEO of CAMINNO, a company building hybrid AI software for engineering and manufacturing optimization in aerospace, defense, and advanced energy. Anika traces her path from a business-focused family into mechanical engineering and a PhD, then into nuclear fusion at Focused Energy. She explains how the disconnect between slow physics simulations and real manufacturing led her to start CAMINNO, using scientific machine learning to run optimizations thousands of times faster than traditional methods. The conversation covers how CAMINNO brings AI into closed loop manufacturing, why trust and validation still matter in high stakes engineering, and how the company has already partnered with national labs like Los Alamos and Sandia. Anika also discusses the challenges of fundraising while delivering active projects, and offers her view on how engineers will work alongside AI a decade from now, not replaced by it but empowered as super engineers.

A fusion founder turned AI CEO explains how simulation software 2,000 times faster is reshaping aerospace and defense engineering.

The questions, and the answers.

You spent your career in deep tech and founded Focused Energy, a nuclear fusion company, in 2021 which is still active. What are you most passionate about today, Focused Energy or CAMINNO?

Honestly both, because my passion is deep tech itself. Focused Energy let me work on bringing the future of energy through nuclear fusion. CAMINNO grew out of noticing huge gaps in engineering, from first principle physics equations all the way to manufacturable products. CAMINNO started by solving the hardest problem, plasma target modeling for fusion, then expanded into aerospace and defense.

What pulled you into deep tech and engineering rather than business or software?

My family was in business and I tried that but found it boring. I thought about becoming a physics and math teacher, then discovered engineering and loved it. I studied mechanical engineering and did my PhD in the field. The more I learned, the more I saw the possibilities and got excited, and it just became my world.

What did fusion teach you about long horizon engineering?

It taught me we need a much better way to bring physicists and engineers together. They have very different mentalities and approaches to system design, and there is constant friction between them based on how each group is trained to think. I believe development in fusion is slowed largely because these two groups work in conflict instead of close collaboration.

What problem did you keep hitting in fusion and aerospace that made CAMINNO necessary?

The disconnect between what you can simulate and how long it takes. Some multiphysics simulations take weeks, and the result still is not close to something manufacturable, which stretches timelines heavily. Scientific machine learning with a physics backbone changes that. Instead of one slow simulation and one test cycle, we can run 2,000 optimization iterations to reach a near optimal design.

In industries where mistakes are catastrophic, how do you get engineers to trust an AI result?

You stay within your engineering realm and always test and verify. We built in basic physics simulations alongside the AI, so if a data point is missing we call the simulation to fill the gap, and every optimization gets double checked. That validation loop brings engineers along by showing them the same answer they would reach, just 2,000 times faster.

Closed loop manufacturing means the system corrects itself mid production. What was the hardest part of making that real?

Most of this credit goes to my co-founder, but the hardest part is getting sensor data into the AI optimizer and back in time, correctly sequenced, so you are not late for the next layer in something like additive manufacturing. Being too late after conditions change can cause a major defect. It now works almost in real time, and we are exploring quantum compute to push speed even further.

Fusion, aerospace, and defense are some of the slowest moving buyers on earth. How did your first customers find you?

We started with a nuclear fusion problem and worked with the Department of Energy, getting grant money to build the initial software. From there we went to trade shows, since most teams already had this simulation bottleneck, making it an easy first pitch. Every team we have worked with since has either sent a letter of capital commitment or signed a three year contract.

If this shift is as big as you think, what does an engineer's job look like in ten years?

Engineers will use AI as a working tool that makes them into super engineers. There is less need for raw programming skill since AI handles that fast, but the engineering mind, the judgment about what a customer needs, the business sense, and the creativity to guide the AI will still matter. That human layer will still be necessary in ten years.