ML Engineer, TTS & Voice AI

Cantina Labs; Greater London; On-site; GBP 148,600 - 163,460; Full time

Benefits offered by this job

Equity

Health insurance

PTO 42 days

Parental leave

401(k)

Free lunch

Job summary

Cantina is seeking a Research / ML Engineer for our Speech Team to build state-of-the-art speech systems end-to-end, from data specs to production inference. You’ll drive the model data eval flywheel for TTS and related tasks, partnering with research, data, and infra to ship reliable, cost-aware models.

You’ll lead small research projects, design experiments, and develop tooling, contributing to safety and responsible AI while pushing the boundaries of voice cloning and controllable TTS.

Qualifications

• Experience with large-scale audio models and data.
• Strong understanding of transformer/diffusion models and audio language modelling.
• Expertise in multi-node distributed training and PyTorch.
• Proven ability to ship production-ready speech models.

Responsibilities

• Architect, implement and fine-tune large-scale speech models.
• Lead small research projects and collaborate on larger goals.
• Design and run experiments to advance model quality.
• Develop dev tooling to boost team productivity.
• Contribute across the stack from low-level optimizations to model design.
• Define data needs and oversee data acquisition, curation and labeling quality.
• Design automated evaluations and monitor model safety and bias.
• Harden training to inference pipelines and manage latency and cost.
• Collaborate with infra to scale training on cloud fleets and productionize models.
• Contribute to safety guardrails and misuse mitigation.

Skills

Large-scale audio models

Transformer architectures

Multi-node distributed training

PyTorch

Production-quality code

Voice cloning / speech control

Publications / open source

Tools

CUDA

Triton

Distributed training tooling

Python

Job description

Cantina is seeking a Research / ML Engineer for our Speech Team to build state-of-the-art speech systems end-to-end, from data specs to production inference. You’ll drive the model data eval flywheel for TTS and related tasks, partnering with research, data, and infra to ship reliable, cost-aware models.

You’ll lead small research projects, design experiments, and develop tooling, contributing to safety and responsible AI while pushing the boundaries of voice cloning and controllable TTS.