Multimodal Infrastructure
Core
Design, develop, and maintain large-scale multimodal data processing, model pretraining/post-training, and inference/serving frameworks.
Role type
Senior IC multimodal infrastructure engineer
Builds
Large-scale multimodal model training and inference systems
Domain
AI/ML infrastructure, distributed systems, multimodal data processing
Deliverable
production ML models
Required skills
distributed data processing infra, CPU/GPU batch processing optimizations, model inference and serving frameworks, image/video/audio data processing, deep learning frameworks, distributed training techniques, model design and optimization, serving frameworks, quantization and compression techniques, distributed inference scaling
Preferred skills
experience in image/video generation and editing, efficient architectures, reinforcement learning training methods, distillation techniques
Technologies
Ray, Spark, PyTorch, Megatron, Deepspeed, vLLM, TensorRT-LLM, SGLang, xDiT, Cache-DiT, Ray Serve, Triton
Responsibilities
Design and maintain multimodal data processing pipelines; Design and maintain model pretraining and post-training frameworks; Design and maintain model inference and serving frameworks; Work with research scientists and product engineers to solve infra-related problems; Lead technical projects and support architectural decisions with data
Seniority
Senior, hands-on IC