Model Research Engineer (m/f/d)

Role overview Research, train, and refine speech-to-speech models that capture human vocal qualities such as emotion, intonation, and natural speech patterns. This role combines applied deep learning research with production engineering: you’ll help determine how models perform in real-time use and build the systems that support fast, reliable iteration.

Responsibilities - Design, train, and evaluate deep learning models, developing capabilities such as emotion recognition. - Integrate model advances into the core architecture. - Optimize inference speed and efficiency for large datasets and real-time workloads. - Write production-ready software that brings research into dependable services. - Build and maintain data processing, training, and evaluation pipelines.

Requirements - Deep expertise in Python and modern deep learning frameworks such as PyTorch or NeMo. - Hands-on experience with large-scale, high-performance ML systems, particularly language modeling for text or audio. - Ability to work in an early-stage codebase and take ownership of technical projects. - Strong software engineering skills alongside research experience.

Nice to have - Experience in a startup or deep-tech environment. - Practical work with audio processing, speech-to-text, text-to-speech, or multimodal models. - Interest in exploring Rust-based inference engines. - German fluency.

Benefits and work setup The role offers remote work, flexible asynchronous collaboration, and access to hardware and cloud infrastructure for training and testing large models. The team describes the position as having room to pursue foundational voice AI research.