Nina Korshunova on Hyperganic, Meshless Simulation and Algorithmic Engineering
CDFAM 23 Speaker Series interview
Nina Korshunova joined Hyperganic when her company’s meshless simulation startup DirectFEM was acquired in early 2022. Since then Nina has taken on the role of Development Manager at Hyperganic, overseeing the Core Platform and the Simulation components.
In this interview, Nina explains the research that led to DirectFEM, how it is now being incorporated into Hyperganic’s offerings, and what it takes to educate designers and engineers on algorithmic engineering.
Nina will also be presenting ‘Physics-Driven Generative Design: The Future of Engineering‘ at CDFAM alongside other experts from engineering, academia and software development. In her presentation she will dig deeper into the mechanics and process of simulation driven and algorithmic engineering.
Can you start by explaining how your research for your PhD at the Technical University of Munich transitioned to the founding DirectFEM, to bring quasi-meshless simulation to the market, which was then acquired by Hyperganic after just over a year?
My PhD started in 2016 with a group at the Chair for Computation in Engineering led by Prof. Ernst Rank at the Technical University of Munich. That is where I worked on the topic “from imaging to numerical characterization: a simulation workflow for additively manufactured products”.
It was quite a challenge for me as it dealt with a lot of concepts that were all new to me; a new simulation approach, additive manufacturing, computed tomography, and more. But I was fortunate to be a part of an incredible group at the Chair, and everyone offered a lot of support. All the researchers in the Chair had worked on and investigated different aspects of a novel quasi-meshless simulation method – the Finite Cell Method.
The work also included the development of the group’s in-house code. This was how I met my co-founders, Laszlo and Davide, who worked alongside me for about 5 years. Together, we pushed the boundaries of science and grew the in-house software.

It was a natural flow of events that led to the three of us deciding to take on the challenge of transforming scientific developments into a commercial product. Looking back at how our entrepreneurial journey started, I must say that if it was not for the encouragement of Prof. Ernst Rank and Dr. Stefan Kollmansberger, we would not have considered starting a company.
I still remember the words of our startup mentor, Antoine Leboyer, the Managing Director of TUM Venture Labs for Software/AI, after looking at our first website. He said: “guys, this is a webpage by three PhDs. I don’t think anyone would buy your product…” And that summed up how my incredible learning journey with DirectFEM started – three PhDs without any knowledge of commercials, finances, business, and product trying to build a startup from scratch.
Gradually, the results of hard work started to show. DirectFEM won first prize in the European Venture Programme, and then was listed as a Top 10 Startups in the first phase of the Business Plan competition. After about half a year, we were accepted into XPRENEURS, the business incubator that is part of a large startup ecosystem from UnternehmerTUM.
This is where our paths first crossed with Lin Kayser, the co-founder of Hyperganic. He gave an inspirational talk on the opening day of the new batch and we later spoke about possible collaborations. He gave us the challenge to perform a structural analysis on the Aerospike Rocket Engine demonstrator, a very complex object generated by Hyperganic.

When we delivered the results in a couple of hours, while varying material properties at every point in the object, Lin was very impressed. He mentioned back then: “this is impossible! Conventional simulation tools cannot handle the amount of detail our algorithms create but you can!” That’s when it really became apparent that we are a perfect match.
Together, DirectFEM and Hyperganic could enact a very big change in engineering by providing seamless integration of simulation and design at its core.
Sounds like quite a rollercoaster experience transitioning from the very solo pursuit of a PhD to running a large team?
I would not say it was ever a solo pursuit. I would even dare to say that the team is what makes things work. Even back in my PhD days, I have never worked alone on my project. Although every researcher worked on separate topics, they were all branching investigations with the same premise at their core.
Together with Philipp Kopp, a researcher from the chair, I was leading the Software Development in the team. In a way, we worked as a very non-traditional research team and had already required a lot of research-based software development processes in place. We had regular standups, coding standards, testing sessions and all these coupled with endless coffee discussions, spontaneous ideation and challenging one anothers’ approaches. Well, the most fun of all was having a regular “Coding Week”, where we put our research aside to work just on cleaning the code or making the software more maintainable.
Only by truly collaborating, can we push the boundaries further.
During the startup time, my co-founders and I brought with us that foundation that was forged during the PhD days. Tackling challenges together, supporting each other at every step of the journey. This was what made everything possible. And now, there is a large team that I rely on to make things work. Well, if I were to be brutally honest, this kind of work would never be possible without a strong team. For me, “running” a team is not the right word, we are co-creating together to generate more value for engineering.

What is the core differentiator of quasi-meshless simulation and what applications does it enable that cannot be addressed by other methods?
A very good question. Quasi-meshless simulation removes one very labor-intensive step that conventional simulation approaches rely on: meshing.
Meshing requires a lot of know-how, many engineering hours, and a lot of frustration from endless iteration at this step.
With our method, we are shifting the effort from the manual and error-prone procedure to a computational step that is fully automatic. In fact, we are helping simulation engineers save around 80% of their time! This, of course, is not all.
The quasi-meshless approach allows us to be geometry-agnostic. I think I would speak to the heart of every simulation engineer who struggled to perform analysis of an object coming from a flawed STL or, even worse, from computed tomography or a point cloud, where there is yet another step of reconstructing the object itself.
We are opening up a whole other world of advantages when dealing with such objects. Our tool does not require a STEP file or a perfectly closed STL – we can deal with any format, without any manual involvement of the engineer. With this, simulation engineers can finally focus on what is truly important and exciting: generating more use cases, and creating more capabilities.
Of course, going a step further, this approach could also account for the material of a bone in a simulation directly from computed tomography, or simulation of an object that comes out of a multi-material printer, where every point in space has different properties. As it is fully automatic, one could connect design and simulation in a seamless feedback loop and grow the geometry based on physical results… And we’re just getting started!

How is this then incorporated, or used with Hyperganic core modeling capabilities?
Indeed, Hyperganic’s Design and Simulation “speak the same language”. This allows for really a lot of possibilities.
Be it a closed feedback loop between design and simulation or traditional product simulation: all of this is possible.
Recently, we have been experimenting with how to direct design with derived physical results. First, an engineer would have to write a few lines of code. Then, one needs nothing more than to press “start” and see where the process brings the design to. Once again, there is no manual interaction.
Can you tell us a little about what you will be addressing with your presentation ‘Physics-Driven Generative Design: The Future of Engineering‘ at CDFAM in NYC?
I would like to mention only a few things about it, because otherwise, it would not be interesting to hear our presentation at CDFAM.
In general, what we ‘ve been noticing more and more at Hyperganic is that when we speak about generative design, additive manufacturing, and physics, there is a large knowledge gap that exists. And I would like to shed some light on it.
The future of engineering is not defined by one person or one organization, but by the community. It is powered by sharing, exploring, and encoding knowledge. It is driven by physics.
Sounds like the challenge, but there are quite a few steps we can take to make a difference in the next generation of engineering.



Without asking for a product roadmap per se, how do you see this design philosophy being realized with Hyperganic?
At Hyperganic, we believe that Algorithmic Engineering is a game-changer. It’s the process of transcribing engineering knowledge into designs through advanced algorithms. With Algorithmic Engineering, we allow engineers to go beyond their imagination and create designs that are as complex and functional as those found in nature.
Our philosophy is about more than just generating complex geometries with code at the modeling stage. We consider all stages of the design process, including designing for specific materials and manufacturing technologies. By making engineering knowledge tangible and accessible through code, engineers can write an algorithm once and reuse it indefinitely or cross-pollinate with objects from other industries.
But we don’t just want to be a “black box”. Our approach is more of a “sandbox”, making our technology and algorithms transparently clear so that engineers can learn from each other. With this mindset, we believe that engineers can unlock the full potential of Algorithmic Engineering and revolutionize the industry.
What do you think are the educational and social barriers to adopting this way of approaching design and engineering once we have all some of the technical hurdles cleared?
When it comes to adopting a new approach to design and engineering, there are definitely some educational and social barriers that we need to overcome. From an educational standpoint, one of the biggest challenges is addressing the knowledge gap between people who are uninitiated to the world of Additive Manufacturing and subject matter experts.
We need to raise awareness of Additive Manufacturing and help grow the industry by bridging this gap. This includes addressing the needs of teenagers who have only meddled with a FDM printer, bachelor students who rely on AM vendors to produce parts for a competition, and many more.
Another key educational barrier is reskilling existing engineers to proficiently think and problem-solve under the paradigm of Algorithmic Engineering. This requires buy-in from companies. To shift from traditional engineering design to a more code-based approach, which can be challenging for some engineers.
Social barriers also play a role in adopting this new approach. We need to reshape the minds of engineers who are used to thinking in terms of what they can see and feel rather than math functions and code. This can be a hurdle, but with the right training and support, engineers can become productive under the paradigm of Algorithmic Engineering.
Perhaps the most important social barrier is changing the mindsets of educators at institutes of higher learning. We need to include Algorithmic Engineering as a potential design philosophy alongside more traditional ways of design, going beyond just designing for Additive Manufacturing. By solving this, we can develop talent capable of bringing our technologies forward.
At Hyperganic, we are working to overcome these barriers with the help of our worldwide network of partners, which includes universities, industry associations, governments, and pioneering startups. With their support, we believe that we can successfully implement Algorithmic Engineering as a transformative approach to design and engineering.
The goal of CDFAM is to bring together the leading thinkers in computational design and advanced manufacturing to share ideas and learn from each other, what do you hope to take away from the event?
I certainly hope to raise awareness of how important this community is. I firmly believe that this is where the change starts.
Furthermore, I would like to hear ideas, share some insights, have exciting discussions, and listen to what other members of this community have to say. This is how and where innovation grows. Only by getting to know each other, sharing experiences, and building this community, can we contribute to advancing engineering. And, as I mentioned time and again, we can’t do it alone. Honestly, I am happy to be part of this and can’t wait to see the progress that we could make together.
To hear more from Nina and other experts in computational design for advanced manufacturing, register to attend CDFAM in NYC, June 14-15 2023