Machine Learning Engineer

Anthropic is hiring a Machine Learning Engineer to design, train, evaluate, and deploy state-of-the-art AI systems that are reliable, honest, and helpful. You will work across the model lifecycle—from data curation and training to optimization and productionization—collaborating with research, product, and infrastructure teams to ship impactful ML features at scale. We value honesty and enthusiasm for doing the work, and we’re looking for builders who can translate cutting-edge research into robust, user-facing capabilities.

• Design, train, and evaluate ML models for language and multimodal tasks with a focus on safety, reliability, and performance
• Own end-to-end model development: data pipeline creation, feature engineering, training runs, evaluation, and iteration
• Develop scalable training and inference workflows (distributed training, quantization, serving, and monitoring)
• Build robust offline and online evaluation frameworks, including automatic metrics and human-in-the-loop assessments
• Optimize models for latency, cost, and throughput while maintaining quality
• Collaborate with research to productionize novel methods (alignment, RLHF/RLAIF, distillation, fine-tuning, retrieval)
• Implement data quality and curation strategies, including red-teaming, bias/robustness checks, and safety guardrails
• Create reliable tooling for experimentation, reproducibility, and observability across the model lifecycle
• Work with product/eng to integrate models into user-facing applications and measure real-world impact
• Document designs, experiment results, and best practices; contribute to a culture of honest, rigorous engineering

• Strong experience training and deploying ML models in production (NLP or multimodal preferred)
• Proficiency in Python and deep learning frameworks (PyTorch or JAX) and common ML tooling (Transformers, Accelerate, Ray, etc.)
• Hands-on experience with distributed training, experiment tracking, and model/version management
• Ability to design reliable evaluation pipelines, including metrics, A/B testing, and human feedback loops
• Experience optimizing inference (serving stacks, quantization/LoRA, batching, caching, GPUs/TPUs)
• Solid understanding of data pipelines: collection, cleaning, labeling, sampling, and dataset governance
• Familiarity with safety, robustness, and alignment techniques (RLHF/RLAIF, red-teaming, constitutional or policy-based methods)
• Comfortable working across infrastructure (cloud, containers, orchestration) to ship ML systems end-to-end
• Clear, honest communication and collaborative problem-solving; enthusiasm for hands-on engineering
• BS/MS/PhD in CS, EE, or related field, or equivalent practical experience