Lead Data Scientist: Recommender Systems & RL

Capital One; New York (NY); On-site; USD 130,000 - 190,000; Full time

Job summary

Capital One seeks a Manager, Data Scientist focused on Advanced Recommenders and Personalization Systems. You will design and deploy state-of-the-art recommender systems using transformers, LLMs, and reinforcement learning across billions of customer records to deliver personalized experiences across marketing, servicing, and digital products.

You will lead a team that builds foundation models, explores multi-modal learning, and optimizes long-term engagement while collaborating with

Qualifications

• Bachelor’s degree in a quantitative field plus 6 years of data analytics experience.
• Experience with large-scale datasets and distributed training.
• Strong programming experience in Python and ML frameworks (PyTorch, TensorFlow).

Responsibilities

• Design and develop state-of-the-art recommender systems using transformers, LLMs, and reinforcement learning.
• Advance personalization through sequence modeling, multi-modal learning, and contextual decision-making.
• Build and fine-tune foundation models trained on large-scale customer interaction data.
• Apply reinforcement learning to optimize long-term user engagement and business outcomes.
• Collaborate with engineering and product teams to deploy scalable, low-latency systems.

Skills

Transformers

LLMs

Reinforcement learning

Python

PyTorch

TensorFlow

Distributed training

Big data

Education

Bachelor’s degree in a quantitative field (Statistics, Economics, OR, Analytics, Mathematics, CS)

Tools

PyTorch

TensorFlow

Job description

Capital One seeks a Manager, Data Scientist focused on Advanced Recommenders and Personalization Systems. You will design and deploy state-of-the-art recommender systems using transformers, LLMs, and reinforcement learning across billions of customer records to deliver personalized experiences across marketing, servicing, and digital products.

You will lead a team that builds foundation models, explores multi-modal learning, and optimizes long-term engagement while collaborating with