Anthropic — Machine Learning Systems Engineer (Infrastructure & Runtime), Horizons

Join Anthropic’s Horizons team to build the foundational infrastructure that powers frontier AI research. You’ll design secure execution environments, high‑throughput data pipelines, and performance‑optimized runtime systems that directly support reinforcement learning and agentic capabilities in Claude models.

What you’ll do

• Design & run high‑performance data pipelines for large‑scale code/LLM datasets with reliability and reproducibility.
• Build secure sandboxed execution environments using technologies like gVisor and Firecracker.
• Develop infrastructure for RL training environments, balancing security with performance.
• Profile, benchmark, and optimize distributed compute to improve utilization and latency.
• Partner with researchers to ship production‑grade systems for AI experimentation.

What you’ll need

• Proficiency in Python (async/concurrency e.g., Trio) and systems programming; strong performance instincts.
• Hands‑on with containers/virtualization and Kubernetes; IaC tools like Terraform or Pulumi.
• Experience building ETL/streaming data systems; secure code execution controls.
• Bonus: Rust and/or C++ for performance‑critical components.

How to stand out

• Share a systems deep‑dive (e.g., sandbox design, kernel/virt choices, perf profiling) with before/after metrics.
• Demonstrate reproducible pipelines (tests, infra‑as‑code) and incident‑ready SLOs for training/inference.
• Showcase collaboration with research partners—turning experimental needs into production runtimes.

Apply at Anthropic →

Updated: 27 Aug 2025

More roles

Even more roles