Researcher, programmer and AI expert with a solid experience in the industry. Currently working as a Senior AI Engineer at Chaptr.AI, developing large language models and AI assistants. Previously, as NLP Team Leader at Voicelab.AI, led the team that designed and trained TRURL, the first ChatGPT alternative in Poland. Successfully defended PhD at Gdańsk University of Technology on detecting and reducing the impact of errors and biases in AI data and models.
With over 2500 citations on Google Scholar, Dr. Mikołajczyk has contributed to numerous research projects and publications in machine learning, focusing on bias detection, explainable AI, and natural language processing. She has led multiple AI for Good initiatives, including HearAI for sign language recognition and DetectWaste for environmental applications.
PhD in Machine Learning, 2017 - 2022
Gdańsk University of Technology
MEng in Control Theory, 2016 - 2017
Gdańsk University of Technology
BSc in Automation Control and Robotics, 2012 - 2016
Gdańsk University of Technology
Professional work and research initiatives
Working with AI assistants, LLMs, LLM-based agents, RAG, and others. Currently working in the book metadata optimization project.
Currently, in contrast to shallow models exploited in the past, most deep learning systems extract features automatically, and to do …
The proposed classify-waste benchmark is a merged collection of publicly available datasets with eight classification labels. The …
Deaf people are affected by many forms of exclusion, especially now in the pandemic world. HearAI aims to build a deep learning …
Speech transcripts generated by Automatic Speech Recognition (ASR) systems typically do not contain any punctuation or capitalization. …
Dataset TinyHero includes 64x64 retro-pixel character. All characters were generated with Universal LPC spritesheet by makrohn. Each …
In the last twenty years the interest of automated skin lesion classification dynamically increased partially because of public …
Sound-Based Bird Classification using Convolutional Neural Networks and Mel-Cepstrum Sepctrograms
Using detection models to localize and classify waste on images and video.
Hackathon. Let's do something for our environment.
ML data collections
The proposed classify-waste benchmark is a merged collection of publicly available datasets with eight classification labels. The …
Speech transcripts generated by Automatic Speech Recognition (ASR) systems typically do not contain any punctuation or capitalization. …
Dataset TinyHero includes 64x64 retro-pixel character. All characters were generated with Universal LPC spritesheet by makrohn. Each …