I'm an MSc Data Science student at Warsaw University of Technology and a Generative AI Engineer working on LLM systems, RAG, agentic workflows, and applied machine learning.
I am interested in building AI systems that are not only technically strong, but also genuinely useful to people - especially in education, communication, accessibility, and human-centered decision support.
-
Co-author of an accepted IJCAI-ECAI 2026 Demo Track paper:
From Surface Learning to Deep Understanding: A Grounded AI Teaching and Learning Assistant for Moodle
arXiv: https://arxiv.org/abs/2605.06963 -
Generative AI Engineer at Evertz, working on production AI systems, agentic workflows, and applied LLM infrastructure.
-
BSc Data Science graduate with distinction from Warsaw University of Technology; currently continuing as an MSc Data Science student.
My technical interests are closely connected to human impact. I am especially interested in AI systems that support learning, communication, and deeper understanding rather than only automating tasks.
This perspective is shaped partly by my volunteering experience, including working with people with severe disabilities and supporting community-focused initiatives. Those experiences taught me how important it is to design technology around real human needs, including people whose abilities, communication styles, or access barriers are often underestimated.
An educational AI system integrated with Moodle, designed to support deeper learning through grounded conversational assistance, quiz generation, and progress-aware learning workflows.
My contributions included Moodle plugin architecture, frontend integration, conversational learning flows, quiz and progress-assessment components, prompt engineering, and end-to-end system integration.
Keywords: educational AI, Moodle, LLMs, RAG, prompt engineering, human-centered AI
Paper: https://arxiv.org/abs/2605.06963
Research-oriented project exploring how vision-language models represent concepts, with a focus on interpretability, concept bottlenecks, and sparse representations.
Keywords: mechanistic interpretability, CLIP, sparse autoencoders, concept bottleneck models, PyTorch
Cloud-based educational RAG pipeline designed around scalable document processing and retrieval workflows.
Keywords: RAG, GCP Cloud Run, Pub/Sub, Docker, vector databases, educational AI
Experimental retrieval-augmented generation pipeline exploring Grover-inspired quantum search ideas.
Keywords: RAG, quantum computing, Grover search, Qiskit, Python
Reinforcement learning project focused on training and evaluating agents in a retro game environment, including multiple experiments and ablation-style comparisons.
Keywords: reinforcement learning, DQN, SARSA, PyTorch, Gymnasium
A natural-language interface for querying weather data using structured SQL generation.
Keywords: natural language to SQL, semantic parsing, SQL, Python
Generative AI and LLMs: RAG, agentic workflows, prompt engineering, MCP, LangChain, LiteLLM, vLLM, Ollama, Hugging Face, OpenAI/Anthropic APIs
Machine Learning: PyTorch, TensorFlow, scikit-learn, NLP, computer vision, reinforcement learning, interpretability, SHAP/XAI
Data and infrastructure: Docker, GCP, Azure, Spark, Kafka, Hadoop, PostgreSQL, Elasticsearch, ChromaDB, FAISS, GitHub Actions
Software engineering: Python, Bash, FastAPI, Streamlit, React, Tailwind CSS, testing and code quality tools
I am gradually cleaning up and documenting my public repositories so they better reflect my current work in AI, machine learning, and production-oriented engineering.
Some older repositories come from university coursework and experiments, while my more recent work includes research and industry projects that cannot always be fully open-sourced.