Senior Data Scientist: Ex

We are a high-growth startup using data and machine learning to revolutionise the insurance industry. You will be joining an established team, working to build products that are fundamentally changing the way insurers see the world, enabling them to move from an assumption based understanding of risk, to an empirical, data-driven view.

About the role:

Our data products, which are used by international insurers, ensure that insurers have access to more data than ever that can be used to dramatically accelerate their learning. To this end, we help them acquire data by extracting information from structured and unstructured documents, through dataset linkage and entity resolution, to using internal and external data to provide insight and prediction (incl. human-assisted ML). Cytora’s Extract team is an interdisciplinary team, and you will be working alongside data engineers, software engineers and other data scientists.

We have a challenging pipeline that requires you to work outside the box and beyond the state of the art. We therefore need you to have a deep understanding of how the technology you’re using works and its limits. We are also a team of builders, so you will be coming up with solutions rather than algorithms - owning your idea, communicating with product, engineering and your coworkers to develop and deploy it.

Your focus will be in information extraction, and you will prototype new techniques and approaches to solve hard problems in document parsing, segmentation, NLP, CV, and text classification / inference.

Requirements

• Proven expertise especially in NLP/NLU/IE, ideally in transformer-based approaches to geometric NER (Layout(X)LM), (visual) document understanding (e.g. DocFormer), and document (visual) question answering (e.g. Donut);
  • 5 years of experience in industry, of which at least 3 with a strong NLP focus
• Excellent understanding of the fundamentals of machine learning and a focus on deep learning
  • Deep understanding of the methods for textual and visual embeddings and fine-tuning pre-trained models;
  • Ability to reason about the choice of a cost function and the appropriate optimizer in different situations
  • Good knowledge of the components common to modern deep learning architectures
  • Proficiency in Pytorch
  • Ideally: Understanding of computational geometry and image (ideally document) rectification, segmentation (e.g. DETR) and labelling
• Experience with ML applications that were deployed and served to customers live; appreciation for label noise, concept drift, class imbalance
• Experience with Python, version control (Git) and Unix based systems; preferably have some hands-on engineering/deployment experience
• Ability to work in a fast moving environment
• Drive to share knowledge with peers and empower your team
• Problem-solving and inquisitive mind and a self-starter attitude
• Ability to draw insights from noisy data when facing open ended problems

Bonus points:

• Excited about engineering
• Experience mentoring more junior data scientists
• Research experience - PhD / equivalent industry experience

Benefits

• Stock options
• Enhanced parental leave
• Private health insurance
• Choice of laptop
• Remote first
• Flexi-working
• £2000 work abroad budget each year
• £1500 Learning and development budget each year
• Company trips

Interview Process

• Video call with Chief Data Scientist (30 mins)
• (optionally) Take home exercise
• Video call with Data Science team (90 mins)