About us

The word rixel is our personal brand. We are freelance deep learning developers and data scientists. Instead of making an incorporated entity with a lot of administration requirements we focus on our ideas and values to use or all time and effort for bringing the best possible solutions.

There are a lot of daily things that can separate or connect people. We try to focus on contributing business relationships rather than making money only. We believe in a world where people have the freedom to live their best lives. There is a three-level measurement method when we develop software: Does this software make someone's life easier? Would we use this software? Does this software provide a contributive business relationship between us and the client? If we have three yes, the project always gets a green sign.

Our most important goal is to build fruitful and living human-software corporations where people can focus on things and activities in which they feel comfortable or where a human is essential. There are fields where an AI or a hardcoded algorithm has advantages over humans. However, humans also have a lot of unique skills that will not be beaten by computers. In an ideal world, people can focus on things that they are the best in. So we try to inspire people around us to have courage and motivation to prosper.

We are two career-jumpers with many skills from various fields. Each former job helps us to be better developers. Different jobs and fields have different mindsets that can be merged into a new way of thinking. In the last years we made business contacts with governmental and private entities around the world. Each past project added something to us. Working on problems of NASA or making base research for a hospital to cure a rare disease needs complex skills which grow from experiences on a wide scale.

Legal experience helps us to see the data without any mislead by prior knowledge. It also helps us to write well organized codes. Furthermore, any coding problem has a legal side. We can translate the technical problem to the language of law and vice versa.

There is no solution without business. As former economists, we understand the importance of business planning and design thinking. Businesses can suffer from inefficiency and wasting resources, like money, process time or GPU memory. A well-optimized code is just the first, but important step on the way of an effective resource-management.

As companies affect each other in the real world, the relation between softwares or parts of code can be competitive or cooperative. We use these skills to build better architecture with parallel programming or implement the business logic in the source code where each function, methods or classes support the optimal running.

Safe and respect-based environment is necessary to have good business connections with colleagues or competitors. We think building new and prospering relations are as important as finishing the job within the deadline.

Teaching skills help us to train models better and understand the world from the view of the models. It is really important, since AI should not be a black box. Models can be enormously complex, but the results and the workflow can be visualized and understood. In our daily life we try to bust these myths and translate the world of AI into a common language.

We really like hackathons. They help us to widen our burdens, learn new things and constantly test our skills and cooperation under pressure. Some years ago it was unimaginable for us to work with Earth Observation (EO) data from satellites or to work with real genomic data. We are problem-oriented developers, we are interested in any topic that needs a solution.

We develop and fine-tune machine or deep learning models. We are eager to use the latest cutting edge solutions like reinforcement learning or private and secure AI training. Thinking out of the box helps us to find a well written short script solution when it reaches the goals better than a robust neural network.

Starting a model training with good fitting hyperparameters is not magic nor a secret knowledge. It is just about training a lot of AI models for various tasks day by day. Practice and innovation always leads to the best solution. The world is full of real problems, and we like challenges and we like to work with AI to solve them.