Staff ML Data Scientist

Role overview Build and deploy large-scale language-model recommender systems, owning the applied machine-learning work from raw customer data through production inference. You’ll work with a science team and customer-facing technical partners to create models that rank and reason over travel and catalog information. The position calls for substantial production ML experience and strong judgment across data preparation, training, evaluation, and deployment.

Responsibilities - Transform structured and unstructured travel and catalog data into narrative and synthetic training documents at tens-of-millions scale. - Decide which features and contextual signals belong in the data and how to represent them for recommendations. - Run distributed model training and fine-tuning, including supervised methods and reinforcement learning. - Own model architecture, ranking components, and co-training or multi-task approaches. - Deploy inference systems with attention to latency and cost, using techniques such as quantization, distillation, and profiling. - Build evaluation sets, run ablations, identify data leakage, and communicate findings to customer and science stakeholders. - Protect customer data, including removing personal information and working within customer environments.

Requirements - Hands-on Spark/PySpark and Hadoop ecosystem experience, including Hive or HDFS, at large document-corpus scale. - Practical experience with SageMaker for data processing, distributed training, and deployment. - Experience fine-tuning language models with LoRA/QLoRA and reinforcement-learning approaches such as GRPO or PPO-style methods. - Strong PyTorch and deep-learning fundamentals, including architectures, training dynamics, and co-training. - Experience with large-scale batch or production LLM inference, such as vLLM or managed batch inference. - Knowledge of AWS data privacy and security practices, including personal-data removal and services or controls such as IAM, KMS, or Lake Formation. - Clear written communication and demonstrable use of AI-assisted development tools.

Nice to have Experience with specialized AWS accelerators, recommender ranking metrics or generative recommenders, embedding quantization or semantic IDs, and customer-facing or embedded consulting.