Member of Technical Staff — Model Optimization and Inference (Experienced)

About Nuance Labs Nuance Labs is building photorealistic, real-time AI avatars with emotional intelligence: a full-duplex audiovisual system that can listen, speak, react, interrupt, and respond like a real person. We're a research company, with PhDs from MIT, UW, Oxford, CMU, and Johns Hopkins, and industry experience from Apple, Meta, Amazon AGI, and Discord. The team is small, the work is real, and the problems are unsolved.

How Nuance Differentiates Most conversational AI avatars today are hacks — a face slapped on a speech-to-speech pipeline, stuck in the uncanny valley: emotionless, mechanical, one-turn-at-a-time. Current systems take 2–5 seconds to respond; natural conversation requires sub-500ms. That's a 10x improvement, and it demands rethinking the entire stack.

That rethinking starts with full-duplex: an AI that listens and speaks simultaneously, perceives emotion in real time, and responds with a face that actually reflects it. It's an extremely hard problem, and we're developing foundation models designed for it from the ground up.

About The Role

We can train a great model. The next problem is making it fast enough to actually use in a real-time conversation — and that gap is enormous. A model that responds in 3 seconds is a demo.

A model that responds in under 500ms is a product. We’re looking for someone who specializes in taking trained models and squeezing every last millisecond out of them. You understand the full stack from model weights to serving infrastructure — quantization, KV cache optimization, kernel-level acceleration, batching strategies — and you know which lever to pull for which problem.

You’ve worked with vLLM, SGLang, or similar frameworks at scale and have strong opinions about where they fall short. This posting is aimed at experienced engineers and researchers who’ve operated at a senior to senior-staff level at big tech, a leading AI lab, or a high-traffic inference team. Everyone at Nuance is MTS — we don’t run title ladders — but we’re hiring people who have already done this work at scale.

Our stack is more complex than a standard LLM deployment: we’re serving a full-duplex multimodal system that must satisfy strict real-time latency constraints. There’s a lot of unsolved optimization work here, and we need someone who finds that genuinely exciting.

What You’Ll Do

Own end-to-end inference optimization across our model stack — LLMs, audio models, and diffusion-based components Implement and tune KV cache strategies for long-context conversations, including eviction policies, compression, and memory-efficient attention Evaluate, deploy, and extend inference serving frameworks (vLLM, SGLang, TensorRT-LLM, etc.) for our specific workloads Profile and benchmark end-to-end latency and throughput; identify and systematically eliminate bottlenecks Build internal tooling that makes optimization work faster and more rigorous — profiling viewers, end-to-end inference test harnesses, and other infrastructure that helps the team move quickly Accelerate diffusion model inference — consistency models, step distillation, caching strategies, and custom kernel optimizations Apply and develop quantization techniques (INT8, INT4, GPTQ, AWQ, and beyond) to reduce memory footprint and increase throughput without meaningfully degrading quality Work closely with research and infrastructure to ensure new models ship with optimized serving from day one

What We’Re Looking For

Significant hands-on experience with LLM inference optimization — you’ve shipped work on KV caching, memory layout, attention kernels, or batching strategies in a production or high-traffic research context Proven proficiency with inference serving frameworks — vLLM, SGLang, TensorRT-LLM, or similar — including going well beyond default configurations and adapting them to non-standard workloads Experience optimizing diffusion model inference (latency reduction, step distillation, caching, or kernel-level work) Strong Python and PyTorch skil