Senior Speech ML Engineer — End-to-End Voice Systems

Cantina; United States; Remote; USD 200,000 - 220,000; Full time

Benefits offered by this job

Equity

Healthcare

PTO 42 days

Parental leave

401(k)

Lifestyle stipend

Lunch on site

One Medical

Job summary

Cantina is seeking a Research / ML Engineer to join the Speech Team to build end-to-end state-of-the-art speech systems, from data specs through production inference. You will drive the model data eval flywheel and collaborate with research, data, and infra to ship fast, reliable, and cost-aware models.

You will design experiments, work with infra and product to ship measurable improvements, and focus on safety and trustworthy AI systems while solving large-scale problems daily.

Qualifications

• Experience with large-scale audio models and intent to ship to production.
• Hands-on diffusion and/or flow-matching transformer experience.
• Experience training audio VAEs, neural codecs, vocoders and related objectives.
• Strong distributed training across multiple nodes/GPUs (FSDP/DeepSpeed).
• Proven ability to ship production-grade, reliable code.

Responsibilities

• Architect and implement large-scale speech models and pre-/post-training stages.
• Design and run experiments to advance model understanding and performance.
• Develop dev tooling to boost team productivity.
• Contribute to full stack from low-level optimizations to model design.
• Define data requirements and oversee data quality and augmentation.

Skills

Large-scale audio models

Diffusion models

Multi-node training

PyTorch

Performance optimization

Open-source contributions

Production-grade code

Tools

PyTorch

CUDA

DeepSpeed

Job description

Cantina is seeking a Research / ML Engineer to join the Speech Team to build end-to-end state-of-the-art speech systems, from data specs through production inference. You will drive the model data eval flywheel and collaborate with research, data, and infra to ship fast, reliable, and cost-aware models.

You will design experiments, work with infra and product to ship measurable improvements, and focus on safety and trustworthy AI systems while solving large-scale problems daily.