Senior Inference Engineer
Remote
Core
Own the inference stack powering multimodal speech-text and text reasoning LLMs for millions of daily real-world voice AI interactions.
Role type
Senior Inference Engineer (Multimodal/Speech)
Builds
Production inference systems for multimodal audio-text workloads and agentic reasoning models
Domain
Conversational AI, Speech-to-Speech, Multimodal LLMs
Deliverable
production ML models
Required skills
CUDA graph optimization, KV cache management, custom kernel development, Python, C++, GPU profiling, scheduler design, batching logic
Preferred skills
Experience modifying inference framework internals, knowledge of memory bandwidth limits
Technologies
A10, A100, H100 GPUs, CUDA, Python, C++
