Research, train, fine-tune, and productionize speech-to-text models across 24+ European languages. Build reproducible experimentation infrastructure, design and run experiments, evaluate architectures and tooling, and own model roadmap from experiment to deployment on an in-house GPU cluster.