Staff Machine Learning Engineer Voice AI

Skills

About the Role

You will own the voice model-serving roadmap for speech-to-text, text-to-speech, and speech-to-speech workloads. You will optimize inference performance, GPU utilization, batching, streaming pipelines, memory management, reliability, and latency. You will build evaluation frameworks, enable emerging audio model architectures, lead model-partner integrations, diagnose performance bottlenecks, influence serving architecture, and define voice fine-tuning capabilities.

Requirements

• 8+ years of machine learning engineering experience
• Production-scale model serving, inference optimization, or machine learning infrastructure experience
• Experience with vLLM, SGLang, TensorRT-LLM, or equivalent serving engines
• Python and PyTorch proficiency
• GPU optimization knowledge, including CUDA kernels, memory hierarchies, and profiling toolchains
• System design experience
• Technical leadership
• Developer tooling product intuition
• Speech and audio machine learning knowledge
• ASR and TTS architecture knowledge
• Audio signal processing
• Audio codec and tokenization familiarity, including SNAC, Encodec, and DAC
• Speech-model training or fine-tuning experience

Responsibilities

• Own the voice inference roadmap for speech-to-text, text-to-speech, and speech-to-speech models
• Architect and implement voice inference systems for latency, throughput, and GPU utilization
• Design serving architectures for serverless and dedicated endpoints
• Build streaming inference pipelines, batching strategies, and memory management
• Build voice-model evaluation frameworks and benchmarks
• Enable support for emerging audio model architectures
• Lead model-partner integrations from implementation through optimization
• Profile, diagnose, and resolve performance bottlenecks
• Influence platform serving architecture for real-time voice APIs
• Define voice-model fine-tuning capabilities

Benefits

• Startup equity
• Health insurance