Kamil Tagowski, PhD
Postdoctoral Researcher & AI Engineer
Department of Artificial Intelligence, Wrocław University of Science and Technology · Wrocław, Poland
Founder & Lead of Finance AI, a research group at Wrocław Tech, within OdysAI.
I work at the seam between research and engineering. My academic background is in graph representation learning and natural language processing, and my day-to-day is building systems that take those ideas out of papers and into tools people rely on.
That has meant co-authoring LEPISZCZE, a NeurIPS 2022 benchmark that standardized how Polish language models are evaluated, and it has meant shipping production legal and tax AI platforms that index millions of documents and serve working researchers. I care about the unglamorous parts that decide whether a model is actually useful: clean datasets, honest evaluation, and software that does not fall over.
More recently that work has grown to include vision and generative models: vision-language systems that read imagery and generative models that create it, like an AI pipeline for generating and editing game maps. I also founded and lead Finance AI, a research group at Wrocław Tech, part of the OdysAI applied-AI initiative.
If you are working on a problem that involves language, graphs, vision, or real-world documents, I would like to hear about it.
Research groups
Department of Artificial Intelligence
Postdoctoral Researcher & AI Engineer
Expertise
Natural Language Processing
Transformer models, retrieval, extraction, and evaluation for real-world and low-resource text, including Polish.
Graph Representation Learning
Embeddings and graph models for entities, documents, and the relationships between them.
Vision & Generative AI
Vision-language models and generative image models: reading and synthesizing visual content, from game-map tiles to document layouts.
Legal & Document AI
Understanding, search, and reasoning over large corpora of legal and administrative documents.
Benchmarks & Evaluation
Designing datasets, metrics, and reproducible harnesses, from human annotation to LLM-as-judge.
Datasets & Pipelines
Versioned data infrastructure with DVC, active learning, and human-in-the-loop curation.
Research Engineering
Turning research prototypes into maintainable systems with FastAPI, PyTorch, and modern tooling.