Remote Production AI Inference Engineer

UMAI Consultancy; United States; Remote; USD 140,000 - 210,000; Full time

Job summary

ElevenLabs is seeking a Research Engineer to join the research team, focusing on deploying and optimizing frontier AI models in production. You will own systems that turn research breakthroughs into real-time products used by millions, ensuring speed, reliability and scalability.

You will work on latency optimization, quantization, distillation, caching, batching, and custom kernels, building tools for researchers to ship models safely and efficiently into production at scale.

Qualifications

• Experience deploying and serving ML models in production, especially latency-sensitive or real-time apps.
• Strong engineering in GPU programming and inference optimization (CUDA, TensorRT, Triton).
• Ability to profile, diagnose, and remove bottlenecks across the serving stack and build tooling to measure it.

Responsibilities

• Deploy state-of-the-art models to production and own the path from research checkpoint to serving infrastructure.
• Optimize inference performance across latency, throughput, and cost using quantization, distillation, KV-cache, batching, and custom kernels.
• Build and tune high-performance serving systems for real-time, streaming workloads where every millisecond counts.
• Create tooling and infrastructure that lets researchers ship new models to production quickly and safely.

Skills

ML deployment

GPU programming

Performance optimization

Tools

CUDA

TensorRT

Triton

vLLM

SGLang

Job description

ElevenLabs is seeking a Research Engineer to join the research team, focusing on deploying and optimizing frontier AI models in production. You will own systems that turn research breakthroughs into real-time products used by millions, ensuring speed, reliability and scalability.

You will work on latency optimization, quantization, distillation, caching, batching, and custom kernels, building tools for researchers to ship models safely and efficiently into production at scale.