ML Infrastructure Engineer

Sygaldry Technologies is building quantum-accelerated AI servers to speed up training and inference for AI. The ML Infrastructure Engineer will build and manage the compute platform for the AI & Algorithms team, ensuring reliable GPU access and efficient workload management across various cloud providers and on-prem GPU servers.

Responsibilities:

• Build compute abstractions that handle the team's diverse workloads: GPU-accelerated simulation, distributed training, high-throughput CPU jobs, and interactive analysis -- across PyTorch, JAX, and scientific computing frameworks
• Stand up experiment tracking and reproducibility infrastructure
• Create developer tooling that makes cloud compute feel local: environment setup, job submission, monitoring, and artifact management
• Scale experiments from single-GPU prototyping to multi-node production runs
• Design multi-provider workload orchestration: route jobs based on cost, availability, and capability
• Manage and optimize spend across cloud providers -- track credit balances, burn rates, and expiration dates
• Configure hybrid local + cloud workflows as on-prem GPU infrastructure comes online
• Coordinate with our infrastructure engineer on cloud administration and security
• Build CI/CD pipelines for research workloads: automated testing, evaluation benchmarks, artifact management
• Create data generation and preprocessing pipelines at the throughput the team's simulators demand
• Set up monitoring, alerting, and cost dashboards that surface problems before researchers hit them