Inference Engineer
Gradium · Paris, FR
Infrastructureonsite
At a glance
Highlights
- Make Gradium's voice AI the fastest and most reliable on the market.
- Own end-to-end inference pipelines shaping latency on every request.
- Partner directly with founders and research to bring new models to prod.
Ideal candidate
Production-obsessed owner with founder mindset who thrives in ambiguity, acts with urgency, and takes pride in ultra-low-latency systems that stay reliable under real load.
Requirements
- Strong C++ and CUDA programming skills with proficiency in PyTorch or JAX
- Experience serving generative AI models in production with vLLM or equivalent
- Track record of measurable latency or throughput improvements
- Take full ownership and act with urgency in ambiguous environments
- Obsessed with reliability and performance under real production load
- Use AI-powered tools daily to improve how you operate
Lab profile
About Gradium
Frontier voice AI company building human-like text-to-speech engine for next-generation voice agents.
What they build
- Text-to-Speech engine for voice agents
- Real-time voice understanding models
- Autonomous voice resolution with human escalation
Known for
- Frontier speech research brought into production
- Genuinely human-sounding voices
- Reliable handling of acronyms, alphanumerics, and cross-lingual names
Key facts
- Spun out of Paris-based Kyutai AI lab
- Small early-stage team
Posted · Last seen