ML Infrastructure Engineer

Key Responsibilities

• Create flexible and performant ML infrastructure
• Design and build systems ML cloud infrastructure to enable massive-scale modeling and analytics
• Support diverse model exploration, hyperparameter optimization, pretraining, fine-tuning, and evaluation processes
• Design and optimize scalable distributed training pipelines, with support for features such model sharding, cross-GPU communication, and real-time training monitoring
• Create, operate, and maintain robust ML platforms and services across the model lifecycle
• Make informed architecture decisions that balance performance, cost, reliability, and scalability
• Build diverse and scalable data platforms
• Design, build, and optimize massive-scale databases and data pipelines for scalable, flexible, and reliable data access
• Explore research-driven, tailored data solutions using existing and simulated data, comparing performance and efficiency across solutions for typical data-access patterns
• Create infrastructure and pipelines for ingesting internal and external datasets with varied shapes, formats, and associated metadata
• Design and assess custom data formats for efficient storage and slicing of high-dimensional time-series data
• Enable efficient data movement, preprocessing, and artifact management for data lineage and modeling reproducibility
• Meet company standards for delivered solutions
• Establish best practices for reliability, observability, reproducibility, and operational excellence across the ML ecosystem
• Make informed and collaborative decisions with domain experts across the software & ML teams
• Foster visibility and reproducibility within the company by maintaining extensive documentation of design decisions, evaluations of viable alternatives for selected solutions, pipeline assessments, etc.
• Support ML R&D operations while preparing for eventual incorporation into product pipelines

Required Qualifications

• Bachelor's degree in Computer Science, Electrical Engineering, or a related technical discipline
• 5+ years of industry experience in software engineering, large-scale data infrastructure, or systems ML
• Extensive proficiency in Python
• Familiarity with PyTorch
• Experience designing, building, and maintaining high-throughput data pipelines for large and diverse datasets
• Experience working with distributed-training frameworks (e.g. FSDP, DeepSpeed, Megatron-LM, Ray, etc.)
• Experience building or optimizing ML training pipelines for transformers or other large neural-network models
• Demonstrated ability to partner closely with research and modeling teams to productionize workflows
• Excellent communication and collaboration skills to work effectively on cross-functional and interdisciplinary teams
• Experience having technical ownership over at least one successfully implemented collaborative project

Preferred Qualifications

• Advanced degree (MS or PhD) in Computer Science, Electrical Engineering, or a related technical discipline
• Proficiency in C++, Go, CUDA, Rust, and/or Java
• Experience in data engineering and systems ML for time-series data
• Deep understanding of the fundamentals of distributed systems, including scalability, fault tolerance, monitoring, observability, scheduling, performance tuning, and resource management
• Experience with cloud-native environments and orchestration (Kubernetes, Docker, etc.)
• Experience scaling foundation-model training infrastructure or multi-cluster computing environments