🛠 Tech Stack

💼 About This Role

You'll lead speech and translation model research for DeepL Voice, shaping the next generation of real-time multilingual communication. Your work will directly impact live speech-to-speech interpretation products used globally. You'll own the full lifecycle from rapid prototyping to production deployment.

🎯 What You'll Do

• Lead hands-on research across ASR, MT, TTS, and speech-to-speech translation.
• Design, train, and optimize large-scale ASR models for multilingual accuracy and low latency.
• Own model delivery lifecycle from prototyping to production deployment.
• Mentor researchers and engineers to elevate model quality and operational excellence.

📋 Requirements

• Deep expertise in speech, audio, or multilingual ML (ASR, MT, TTS, or end-to-end ST).
• Hands-on builder who trains models, runs experiments, and integrates ML systems into production.
• Experience shipping ML models to production and maintaining them at scale.
• Strong coding in Python, PyTorch/JAX, and audio processing libraries.

✨ Nice to Have

• Experience with streaming inference and ultra-low-latency systems.
• Published research at top conferences (ICASSP, Interspeech, NeurIPS, etc.).

🎁 Benefits & Perks

• 🏖️ Flexible hybrid work model (Munich office + remote).
• 💰 Competitive equity and performance bonuses.
• 📚 Learning budget for conferences, courses, and books.
• 🏥 Comprehensive health insurance and wellness programs.