SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
voize is an AI company building AI companions that automate digital workflows for frontline workers, enabling them to focus on people and care. Backed by $50M Series A funding (Balderton Capital, HV Capital, Y Combinator), the company serves 2,000+ facilities with 200,000+ active users.
You will own the full lifecycle of deep learning models for speech recognition—from design and training through evaluation, deployment, and optimization. Your models directly power the company's applications and impact end-users in the field. You'll improve a diverse family of speech models across languages and products, engineering fast inference for both mobile devices and hosted environments.
Your daily work includes:
- Design, training, evaluation, and deployment of deep learning models for speech recognition
- End-to-end ownership of the ML process: concept exploration, productionization, maintenance, monitoring, and optimization
- Improving live speech models across languages and products
- Engineering efficient inference on mobile and hosted infrastructure
- Shipping new features to production regularly with direct user feedback
voize offers a hybrid setup based in Berlin with flexibility to work remotely across Germany. The culture emphasizes collaboration, rapid growth, and meaningful impact. Benefits include 30 vacation days plus birthday off, Germany Transport Ticket, Urban Sports Club membership, regular company off-sites, learning platform access (Blinkist, Audible), free language courses, stock options, and flexible working hours.
REQUIREMENTS:
- Several years of hands-on experience in Deep Learning for speech recognition, including developing and optimizing ASR (Automatic Speech Recognition) systems in production (not academic research alone)
- Strong foundation in STT (Speech-to-Text) system development with focus on real-world applications
- Proven ability to own the ML process end-to-end: from concept and exploration to model productionization, maintenance, monitoring, and optimization
- Shipped ML models to production using Python and PyTorch
- Strong software engineering practices: clean code, tests, guardrails
