Job Description

We're on a mission to protect voice data in the age of AI — tackling challenges like voice anonymization, deepfake detection, and adversarial robustness. Our real-time speech models have already processed over a million minutes in production, powering applications that demand privacy, security, and performance.

As an MLOps Engineer (Speech-Focused), you’ll play a key role in scaling our model deployment and data workflows. From managing multi-language speech datasets to deploying PyTorch models in production, your work will directly shape the backbone of our research and product infrastructure. We’re a small, fast-moving team, and you’ll have space to take ownership and grow.

Responsibilities

• Design, build, and maintain infrastructure for deploying PyTorch models at scale.
• Prepare, clean, and manage datasets across multiple languages and tasks.
• Build and optimize training and evaluation pipelines for speech models.
• Automate recurring processes using Docker and internal tools.
• Evaluate model quality and monitor performance metrics across tasks.

Requirements

• Strong experience with Python and PyTorch in real-world ML codebases.
• Comfortable with Docker and solid software engineering practices.
• Able to work independently, take initiative, and follow through on complex tasks.
• Background or interest in speech/audio processing.
• Open-minded and adaptable — eager to learn and experiment with new tools.

Nice to Have

• Experience with ML evaluation frameworks or experiment tracking tools (e.g., W&B).
• Exposure to model optimization or cloud deployment.
• Hands-on experience working with audio or speech datasets.

What We Offer

• A remote-first, flexible environment (160 hours/month) within EU time zones.
• Competitive salary and equity (ask us for details).
• The chance to shape foundational infrastructure at a small but impactful company.
• 3–4 in-person hackathons per year for collaboration, deep work, and some fun.
• Autonomy in how you approach your work — we trust you to get things done.

We value clear communication, curiosity, and a practical mindset. There’s room to grow into research, modeling, or other directions depending on your interests. If you’re excited to build systems that make cutting-edge speech models production-ready — we’d love to hear from you.