Different environments, different objects, a new way of creating
At New Science Lab, we develop AI-powered technology that accelerates the design and optimization of object geometries operating across different physical environments — wherever shape determines behavior, efficiency, and operational cost.
HullGen — Our First Solution
HullGen is our first product: an AI system built for generating and optimizing hull shapes for marine vessels, including autonomous and specialized dual-use platforms. With variant generation measured in seconds, it dramatically accelerates the transition from functional requirements to comparable hull variants, significantly reducing costly CFD iterations in the design process.
The system is currently at the level of a working Proof of Concept at TRL 3.
We are looking for partners for technical validation, design pilots, and R&D collaboration.
Early-stage hull design is too slow, too costly, and limits the number of variants explored.
Conventional workflows require multiple iterations and expensive CFD analyses. As a result, design teams compare too few variants, and poor decisions only surface at later, costlier stages of the project.
For both civilian and military marine vessels, this has direct consequences. Hull geometry affects drag, energy consumption, range, stability, and vessel behavior in a specific mission profile.
How HullGen Works
The user defines a subset of 45 parameters describing the vessel — e.g. dimensions, payload, operating conditions — and sets optimization objectives.
The system creates geometry variants within a structured parametric space and discards solutions that do not meet the defined constraints.
Each generated variant is described by the full set of 45 defining parameters and 138 output parameters characterizing hull behavior, ready for further analysis and evaluation.
The user receives a package: 3D hull geometry (STL or STP file), run log, input and output parameters, and a seed identifier for full reproducibility. The geometry is ready for import into CFD tools and CAD environments.
Where We Are Today
HullGen is currently at TRL 3 — a working Proof of Concept in a laboratory environment. It is not yet a deployment-ready product. The current version demonstrates that it is possible to move from selected input parameters to generated and comparable hull variants in seconds.
We present a tool that can shorten the early design exploration phase and prepare better input for further analyses. HullGen does not replace the full design workflow, a complete CAD/CFD environment, final CFD analyses, or the end validation required before deployment.
Current Model Validation
At this stage, validation means four concrete things for us:
the same parameters lead to the same result within defined tolerance limits
every run is logged and reproducible
the system does not promote physically unjustified solutions
results are benchmarked against reference data within a clearly defined scope of application
Upcoming Work
Goal: pilots with partners and expanding validation to cases closer to real-world projects — transition to TRL 4–5.
Where HullGen Delivers the Greatest Value
HullGen's area of application is the design and optimization of hull shapes for marine vessels.
HullGen can support shipyard design teams and R&D teams in moving through a larger number of shape variants faster, before more expensive analyses are triggered. The result is a platform better matched to a specific task, with reduced design time and lower cost per iteration.
For the end user, a better-matched hull for specific operating conditions translates to lower energy consumption, greater range, stability, and mission capability.
Application areas
See HullGen in Action
You can get acquainted with the current state of the HullGen technology demonstrator. The demo shows how the tool works, the optimization logic, and the form of output — not a final enterprise product.
The current demo is designed to show the technological direction and process architecture. Not all elements of full validation and high-fidelity simulations are yet integrated. Some current predictions are Proof of Concept in nature and will be developed toward greater reliance on additional data and more advanced simulations.
During the presentation we show:
We also discuss the current state of the solution, the scope of meaningful use, and our development plans.
HullGen Is Our First Product, Not the Limit of Our Technology
HullGen is the first practical application of the technology developed by New Science Lab. We start with marine vessel hulls because this is an area where geometry directly affects drag, range, energy consumption, stability, and operational utility.
It is also a good first validation market: the problem is concrete, costly, and well-measurable. At the same time, the method itself has greater potential.
The Broader Ambition of New Science Lab
We are building technology that supports the generation, assessment, and optimization of shapes for objects operating in different physical environments. The emerging product engine, model training methodology, and validation approach are universal across many classes of shapes and environments.
After surface vessels, we plan analogous products for:
Our goal is not to build a single geometry configurator. We are building the foundation for a broader platform combining AI, engineering knowledge, physical models, and design process control.
Team
An interdisciplinary team combining artificial intelligence, modeling and simulation, marine engineering, product development, and deep-tech commercialization experience.
Sławomir Dudek
Entrepreneur and manager with over 25 years of experience building, developing, and restructuring businesses at the intersection of technology, e-commerce, healthcare, and services. Combines a technical background — applied mathematics and technical physics — with management experience gained at Gillette Poland, Pelion, KPMG Advisory, Delfarma, Cefarma Białystok, and clinika.pl. At HullGen, responsible for business strategy, commercialization, partnership development, and preparing the project for financing and deployment.
Tomasz Pochylski
Entrepreneur and technology project operator. Co-founder and former CEO of Bitfold — a hardware deep-tech project developed by a team of around 30 people, which raised over US$5.5 million in public funding and comparable private capital. Responsible for operational project management, execution structure, and the transition from prototype to a structured product.
Krzysztof Witek
Engineer with experience in ocean technology, product development, deployment, and working with industrial partners in the marine market. CEO of Raiton, where he develops projects related to hydrogen technologies, including autonomous watercraft. Responsible for relations with technology partners, research institutions, and investors, and brings a practical perspective on the market and HullGen applications.
Prof. Zbisław Tabor, PhD, DSc
Professor at AGH University of Science and Technology in Kraków. AI researcher and practitioner with experience developing algorithms for industrial and medical applications. Specializes in machine learning, deep learning, computer vision, and model interpretability. Combines research and deployment competencies. At HullGen, responsible for AI layer development, research direction, and technical coherence of the solution.
Michał Sikorski
AI and robotics engineer with experience building machine learning models and applying them in technical and robotic systems. Works with Python, TensorFlow, PyTorch, OpenCV, and ROS. At HullGen, responsible for developing the implementation and algorithmic layer of the system, supporting the transition from model concept to working prototype.
Maksym Prykhodko
Developer focused on Python and ML system integration. Builds systems where strict engineering rules meet generative AI — with experience in time-series data processing, signal filtering, and backend data analysis using PyTorch and FastAPI. In the project, responsible for the software layer enforcing design standards in the hull model.
The team is supported by experts in CFD, marine engineering, and machine learning, engaged at the stages of technical validation, selection of reference methods, and preparation of pilots.
Who We Are Looking For
We are at an early but documented stage of development. Our next steps are the transition from a working Proof of Concept to the first credible version of the product. We are looking for partners for collaborative development and for validating the technology in a real design and operational context.
Let's see if HullGen can shorten your path to a better design.
We welcome contact in any of the following areas of potential collaboration:
Pilot partners
We welcome meetings with organizations ready to test HullGen on a specific use case — particularly in the area of autonomous, specialized, and marine applications requiring faster design iteration.
Data and validation cases
We are looking for access to data, design scenarios, and reference cases that will allow us to better validate the product and develop it toward real engineering utility.
Technology and research partners
We are open to collaboration with research teams, simulation experts, and industry partners who can support the development of the engineering knowledge layer, validation, and physical assessment.
Investors and deep-tech / dual-use programs
We engage with investors and programs supporting the development of deep-tech and dual-use technologies — including innovation programs of alliances and European organizations supporting technologies with potential for civilian and defense applications. We are at TRL 3 and seeking funding to accelerate the transition to TRL 4–5 through pilots in real design scenarios.
Contact