Member of Technical Staff - Pretraining Infrastructure | AI Foundation Models

Member of Technical Staff - Pretraining Infrastructure | AI Foundation Models Compensation: $300,000+ Base with Competitive Equity I'm hiring on behalf of a well-funded AI startup that's redefining the future of conversational artificial intelligence. Backed by $60M in funding, this company is building next-generation multimodal foundation models capable of understanding and responding through audio, video, and language simultaneously. The engineering team consists of researchers and industry experts from some of the world's most prestigious institutions and technology companies, including MIT, Carnegie Mellon University (CMU), the University of Oxford, the University of Washington, Johns Hopkins University, Apple, Meta, Amazon AGI, and Discord. Unlike conventional conversational AI systems that rely on sequential interactions, they're developing real-time full-duplex AI, enabling systems to listen, speak, understand emotions, interrupt naturally, and respond in under 500 milliseconds—bringing human-like conversations closer than ever before. If you're passionate about solving large-scale infrastructure challenges and building the foundation behind the next generation of AI, this is an exceptional opportunity. Key Responsibilities Own the end-to-end distributed training infrastructure for multimodal foundation model pretraining. Design, build, and scale distributed training systems across large GPU clusters. Develop and maintain the complete training runtime, including: Job orchestration Distributed execution Checkpointing Fault tolerance Recovery Monitoring Debugging long-running training workloads Optimize large-scale model training by improving: GPU communication CUDA performance Data throughput Parallelism strategies End-to-end training efficiency Build infrastructure supporting multimodal workloads, including: High-throughput audio, video, and text data pipelines Temporal alignment Variable sequence processing Partner closely with AI researchers and engineering teams to evolve infrastructure alongside new model architectures, hardware platforms, and research breakthroughs. Requirements Experience building or operating distributed training infrastructure for large-scale AI models. Experience training workloads across hundreds to thousands of GPUs. Strong expertise in: Distributed Systems GPU Computing CUDA NCCL High-Performance Computing (HPC) Parallel Computing Hands-on experience with distributed training frameworks such as: PyTorch Megatron DeepSpeed PyTorch FSDP Experience building or modifying ML training runtimes from the ground up. Experience with multimodal AI involving audio, video, and language is highly preferred. Open-source contributions to ML infrastructure projects (SGLang, vLLM, or similar) are a strong plus. BS/MS in Computer Science, Electrical Engineering, or a related technical discipline. Must be located in or willing to relocate to Seattle. Why Join? Join a rapidly growing AI startup backed by $60M in funding. Work alongside world‑class researchers and engineers from globally renowned universities and leading technology companies. Build greenfield infrastructure from 0 to 1 with end-to-end ownership. Contribute to the development of next-generation multimodal AI foundation models. Competitive Compensation & Equity Package. #J-18808-Ljbffr Carnaby Fox