Data Scientist / Deep Learning Practitioner

Capital One Capital One · Banking · London, United Kingdom +1

Develops deep learning models for underwriting and consults on LLM-powered product development in a fintech company.

What you'd actually do

  1. Develop new deep learning approaches to advance our current underwriting models, which form the heart of our lending business.
  2. Apply these to new types of (multi-modal) data in order to stay at the forefront of innovation.
  3. Provide consultancy to our tech and product partners, to help design, develop and launch products powered by Large Language Models (LLMs). This collaboration will help provide seamless experiences for our customers and associates.
  4. Use a combination of business acumen, coding and statistical skills to navigate large amounts of data and extract actionable solutions.
  5. Work cross-functionally on projects that support key business initiatives and drive sustainable growth.

Skills

Required

  • Experience developing and deploying deep learning models, particularly for sequential data (e.g. time series, language) using techniques such as LSTMs or transformers.
  • Hands-on experience with modern Machine/Deep Learning frameworks such as PyTorch, TensorFlow, or Hugging Face Transformers.
  • Familiarity with both pre-training and fine-tuning of large-scale models
  • Experience working with structured and unstructured data, such as text, logs, or time series and tokenisation techniques.
  • A strong understanding of probability, statistics, machine learning and familiarity with large data set manipulation.
  • Experience in producing reliable and maintainable code in Python, with an ability to adapt to new languages and technologies.
  • Ability to communicate findings to a diverse business focused audience, influencing others in both verbal and written form.

Nice to have

  • A drive for continued learning through an internal and external focus, in order to develop enterprise and industry leading solutions.

What the JD emphasized

  • deep learning models
  • underwriting models
  • multi-modal data
  • Large Language Models (LLMs)
  • deep learning
  • sequential data
  • pre-training and fine-tuning
  • structured and unstructured data
  • Python

Other signals

  • developing proprietary deep learning models
  • advancing underwriting models
  • applying to new types of (multi-modal) data
  • consultancy to tech and product partners
  • design, develop and launch products powered by Large Language Models (LLMs)