About the role
Responsibilities:
1. Data to training sets: turn unstructured and structured traveler and catalog data into
narrative text documents and synthetic data, at a scale of tens of millions of
documents.
2. Feature and context engineering: decide what information goes into each document
and how to express it, such as market-relative price, tier and trip context.
3. Large-scale distributed training on HyperPod.
4. Fine-tuning: LoRA/QLoRA adapters, supervised training, and reinforcement learning
for ranking and reasoning.
5. Deep learning modeling: model architecture, ranking heads, co-training and
multi-task setups.
6. Inference and deployment: low-latency, cost-efficient serving, including quantization,
distillation and profiling.
7. Evaluation and error analysis: build evaluation sets, run ablations and catch data
Leakage.
8. Data privacy: handle customer data safely, including removing per
Required skills
• Spark (PySpark) and the Hadoop ecosystem (Hive, HDFS) on corpora of tens of millions
of documents.
• LLM fine-tuning (LoRA/QLoRA) and reinforcement learning (e.g. GRPO/PPO-style
methods).
• Strong deep learning fundamentals in PyTorch: architectures, training dynamics,
Co-training.
• Large-scale batch LLM inference and production inference (vLLM, Bedrock
batch).
• Data privacy and security on AWS: personal-data removal, IAM, KMS, Lake Formation.
• Clear written communication with customer and science stakeholders.
Preferred skills
• AWS Trainium / Inferentia (Neuron SDK).
• Recommender systems: ranking metrics (NDCG, MRR), sequential and generative
recommenders.
• Embedding quantization or semantic IDs.
• Experience in a customer-facing or embedded role.
Requirements
- • Spark (PySpark) and the Hadoop ecosystem (Hive, HDFS) on corpora of tens of millions
- of documents.
- • LLM fine-tuning (LoRA/QLoRA) and reinforcement learning (e.g. GRPO/PPO-style
- methods).
- • Strong deep learning fundamentals in PyTorch: architectures, training dynamics,