Jakub Raczyński

- design & research on the NLU engine for TV voice control - implementation of the scoring system for VoD platform - enhancement of recommendation system for VoD platform, most notably employing a LLM and implementing computation of new metrics, which allow monitoring of the system performance - research on recommendation algorithms for STBs - data wrangling, including advanced, deep learning based NLP techniques - developing data processing pipelines, based on Kafka, Protocol Buffers, Apache Airflow and MongoDB - serving models / development of REST APIs in Python/Flask

- Kubernetes (OpenShift) cluster administration, - Microservices/cluster monitoring with Prometheus/Grafana and ELK stack, - Infrastructure management with Terraform, - Proxmox VE administration, - Gitlab CI pipelines maintenance & development, - Java 11 & Kotlin development

Research internship in Barcelona Supercomputing Center on distributed deep learning, as a part of PRACE Summer Of HPC programme. Supervised by Albert Njoroge Kahira and Leonardo Bautista Gomez.

Graduated with distinction, defended thesis "Explaining neural recommendation systems in natural language", focused on improving state of the art deep neural recommenders explaining their decisions with natural language. Thesis was awarded by City of Poznan in the Contest For the Best Master's Thesis in year 2022/2023