About Me
Greetings! I am a Data Scientist with a strong Software Development background. I have a great passion for machine learning and artificial intelligence, especially for Large Language Models. As a member of agile teams, I have contributed to the entire software lifecycle and participated in cutting-edge research initiatives. More info below:
From idea to product
I am working in a data science team, developing AI products using state-of-the-art technologies. Currently, I am fully immersed in the latest advancements of Large Language Models such as LLaMa, LangChain, agents, information retrieval, chatbots, and various other AI tools.
Aside from the LLMs, I have built machine learning pipelines (DVC, Poetry, Docker, Git) for denoising models, including diffusion ones in PyTorch, web crawlers and APIs with MongoDB and Flask.
PROFINET Data Analysis
With my previous team, we developed, evaluated and optimised ML algorithms for PROFINET diagnostics. We also collected, pre-processed and labelled data. We tested a number of ML algorithms: anomaly detection, autoencoders, time series analysis, cluster analysis and many others.
Our team has developed an XGBoost classifier capable of detecting the root causes of PROFINET network failures. The model was trained in Python and converted to ONNX for integration into a powerful network monitoring application.
A low-dimensional visualisation of the data is shown in the image on the left.
Genetic Algorithms
During my studies, I developed and evaluated the possibility of automating the placement of electronic components on a printed circuit board using evolutionary algorithms. The genetic algorithm optimised two criteria: the total length of the connections and the heat dissipation. The open source project was written in C# using Unity3D.
An example of a generated 3D model is shown in the image.
GitHub RepoCell Laboratory
A simple click-to-feed Android game written in C# using Unity3D. We made it with my friend as a pet project. It works smoothly even with a LOT of cells making decisions.
The game is published on Google Play.
Google Play pageScientific Work
While working on PROFINET projects, we've carried out several research projects in collaboration with universities.
«First Steps on Incident Detection for Field Bus Systems», 11th Mittweida Workshop on Computational Intelligence (MiWoCI), 2019
«Neural Network Data Processing For Analysis of the Industrial Networks Parameters», Artificial Intelligence and Decision Making, 2020