Research Engineer, LLMs

Build and scale systems for training and deploying large language models for video analysis.

• •Mirage is seeking a Research Engineer to build and scale systems for training and deploying large language models for multimodal creative tasks, specifically focusing on video analysis pipelines.
• •Key Responsibilities Build pipelines for large-scale LLM training and fine-tuning, optimizing data loaders for high-bandwidth multimodal datasets Optimize training and inference performance Implement distributed systems for scalable experimentation Improve latency, throughput, and cost efficiency for complex multimodal inputs and structured generation Support deployment of LLM systems into production environments Develop tools for evaluation, monitoring, and iteration Requirements 2+ years of professional industry experience Strong experience with large-scale ML systems Proficiency in PyTorch, CUDA, Triton, and distributed training Experience optimizing LLM training or inference Strong systems intuition (memory, compute, throughput tradeoffs) Ability to bridge research and production Strong software engineering and debugging skills