Machine Learning engineer dedicated to writing high quality code that speaks for itself.

🤖 GenAI engineer @ Accenture
🔬 Former president of AI Science Society "Golem"
💼 Former ML engineer and researcher @ AI division of WUT
🎓 Graduated as BEng in CS and ML @ WUT
🎓 ML MEng student @ WUT

• languages: PythonC++
• machine learning: pytorchscikit-learnwandb.ai
• GenAI: langchainRAG
• multi-agent systems: crewai
• data science: numpypandas
• devops: dockerCI/CDunit testingairflow
• hobby: arduino

machine learningpythonCI/CDdocker

Optilab is a Python framework designed to streamline the prototyping and benchmarking of black-box optimization algorithms. It offers a comprehensive suite of tools for comparative analysis and results reporting, including statistical tests and visualizations like ECDF and convergence plots. Optilab is readily accessible via PyPI and Docker Hub, with comprehensive documentation hosted on Read the Docs. The project utilizes continuous deployment pipelines to ensure timely delivery of new versions upon each release.

GenAIRAGlangchainstreamlit

Don't want to read a book? Chat about it with this AI chatbot instead. This GenAI RAG application allows user to talk to a virtual assistant about a given book, allowing them to understand it better and quicker. With a user-friendly GUI made in streamlit, anybody can use it without any technical knowledge. Booktalk comes in handy when you have a boring school reading to do, nobody to discuss your favourite books with, or when you just need a little reminder of the plot.

pytorchmachine learningwandb.aipython

This project investigated the application of the CMA-ES metaheuristic for training neural networks. While deep learning traditionally relies on gradient-based optimization methods like SGD and Adam, CMA-ES represents a state-of-the-art black-box optimization algorithm known for its efficiency and performance. This research compared the performance of CMA-ES against traditional gradient-based approaches in the context of training neural networks, analyzing the results to evaluate the potential of CMA-ES in this domain.

machine learninglangchainllmpython

Codemate is a command-line interface (CLI) coding assistant designed to alleviate mundane and repetitive tasks in software development. Leveraging large language models (LLMs) such as those powering ChatGPT and GitHub Copilot, Codemate provides developers with tools to efficiently document code, clarify complex sections of source code, and generate new code based on specific requirements.

machine learningpythontensorflow

Computer Vision program that matches image with the most similar YuGiOh! card.

C++unit tests

C++ library for non-linear dialogues in roleplaying games.

arduinoC++

Arduino device to navigate the computer with hand gestures.

mutli-agent systemscrewaillmpython

A multi-agent LLM system to kickstart a new software development project.

pytorchpython

An implementation of Lion optimizer discovered by Google Brain.