I'm Gabriel Bretschner, a Research Scientist at Lilt Inc. in Berlin. My work spans the machine translation stack, from model training to evaluation and benchmarking.
I studied at RWTH Aachen, where I completed both my Bachelor's and Master's degrees in computer science. I started out working on machine translation model training and on automating training and data pipelines. Today my focus is on measuring and evaluating these systems: quality estimation for machine translation in real-world localisation, and building benchmarks that probe how large language models and coding agents behave across languages, domains and cultural contexts.
You can find a list of my research papers and publications on the publications page.