Member of Technical Staff - Multimodal Understanding
Core
Building superhuman multimodal intelligence systems that see, hear, reason about, and interact with the world in real time across image, video, audio, and text modalities.
Role type
Senior IC multimodal machine learning engineer (full-stack)
Builds
Frontier multimodal AI models, distributed training/inference systems, data pipelines, and real-time interactive human-AI collaboration tools.
Domain
Artificial Intelligence / Multimodal Learning / Large-Scale Distributed Systems
Deliverable
production ML models
Required skills
Multimodal pre-training/post-training, Python, large-scale distributed ML systems, data pipeline design, evaluation design and benchmarking, RL techniques
Preferred skills
Rust/C++, large-scale orchestration tools, full-stack tooling, scaling laws and tokenizers
Technologies
Python, JAX, PyTorch, XLA, Spark, Ray, Kubernetes
Responsibilities
Design and optimize large-scale distributed systems for multimodal pre-training, post-training, and inference; Develop high-throughput pipelines for multimodal data acquisition and management; Advance multimodal capabilities including world modeling, reasoning, and real-time video processing; Drive data quality and curation for trillion-parameter models; Create evaluation frameworks and reward models; Innovate on algorithms and hardware/software co-design; Build research tooling and full-stack applications for rapid iteration.
Seniority
Senior, hands-on IC