Marek Kapera

I'm currently working on Apple Maps as a Machine Learning Engineer, following production AI/ML work for media planning at Kinesso and applied AI research at RNRS Solutions.

At Kinesso, I developed JAX and NumPyro Bayesian marketing mix models, designed causal experiments for attribution, and built agentic systems using LangGraph, LangChain, RAG, and specialized tools. I also contributed FastAPI and Dagster components, testing, evaluation, observability, and architecture proposals.

At RNRS Solutions, I took an ambiguous concept through to a deployed Google Cloud system, setting technical direction and coordinating junior contributors. I designed data collection, extraction, scoring, evaluation, and production services, including PyTorch classifiers and LLM-based extraction components.

Alongside industry work, I conduct ML/AI and quantitative research at the Polish Academy of Sciences on beliefs, preferences, uncertainty, learning, and choice behavior. My work spans causal and counterfactual methods, transformer architectures for structured belief systems, and NLP and graph pipelines for misinformation and narrative data.