Data Scientist - Fraud

Equifax; Greater London; Hybrid; GBP 90,000 - 130,000; Full time

Benefits offered by this job

In-office collaboration Mon-Thu

Friday work from home

Job summary

Equifax UK Data & Analytics invites you to develop market-leading fraud scores and models using cloud-based technologies. You will work with a global team to build AI/ML solutions and deliver high-impact analytical products.

You’ll apply traditional and cutting-edge techniques, create new features, and collaborate with stakeholders to deploy scores with governance. The role follows a four-day in-office week (Mon-Thu) with Friday flexibility.

Qualifications

• Experience using credit bureau data in statistical fraud models.
• Extensive experience developing statistical models and scores in fraud.
• Highly numerate with a relevant degree 2:1 or above.
• Day-to-day use of Python, SAS, BigQuery, Jupyter Notebook and SQL.
• Experience developing regression models and scores (e.g., logistic).
• Exposure to large/complex datasets and analytical platforms.

Responsibilities

• Develop market leading statistical fraud models and pseudo-models using diverse data sources.
• Apply traditional and AI/ML techniques (logistic, gradient boosting, random forests, clustering).
• Leverage cloud based tools to support test-and-learn champion/challenger approaches.
• Create new features to improve predictive power and analytical products.
• Investigate segmentation to maximize model performance.
• Collaborate with internal stakeholders to deliver new scores and products.
• Share best practices with Data Scientists globally and adhere to policies.
• Produce documentation for peer review and Model Risk Management.

Skills

Python

SAS

BigQuery

Jupyter Notebook

SQL

Logistic regression

Statistical modeling

ML/AI techniques

Fraud modeling

Communication

Education

Bachelor's degree in a numerate field

Tools

G-Suite

Google Cloud Platform (GCP)

Vertex AI

Job description

Come join the Equifax UK Data & Analytics team and develop market leading fraud scores, models and analytical solutions using the latest cloud based technologies, techniques and tooling. Be part of a growing and diverse team tasked with creating the next generation of statistical models, machine learning algorithms and AI based products and services.

We believe great things happen when teams connect. Our schedule is built around 4 days of high-impact, in-office collaboration (Monday–Thursday), paired with Friday Flexibility to wrap up your week remotely.

What You’ll Do

• Develop market leading statistical fraud models and pseudo-models for the Equifax UK business e.g. consumer fraud models, commercial fraud models, synthetic fraud, payment intent models, and loan stacking indicators and models Utilise and blend all available data sources to ensure all fraud models are highly predictive and market leading - consumer bureau data, postcode insights, open source data,transactional insights & fraud indicators from partnerships
• Apply traditional model development techniques and approaches (e.g. logistic regression) as well as adopting new AI/ML techniques such as gradient boosting, random forests, clustering, etc.
• Utilise the latest cloud based technologies and analytical tooling to support test-and-learn champion/challenger approaches towards new fraud scores and models
• Consider the creation of new characteristics/attributes to improve and maximise the predictive power of the model(s) as well as enhancing and creating other analytical products and services. This can be done through expert approaches or through AI approaches (e.g. transformers)
• Investigate and apply appropriate segmentation and sub-populations to maximise overall model & score performance
• Liaise with internal stakeholders in a timely and effective manner towards the delivery of new scores and products e.g. Product Managers, Model Risk Management, Pre-Sales, Consultancy, Compliance, Technology, Data & Analytics etc.
• Work closely with Data Scientists and analytical teams globally across the Equifax business to share best practice and learnings
• Adhere to relevant model development policies and procedures towards a successful deployment of scores and models e.g. sample design, good/bad definitions, rationale for exclusions, model development, model validation, implementation, etc.
• Produce suitable documentation alongside the statistical development of the model(s) to support peer review and Model Risk Management processes

What Experience You Need

• Experience using credit bureau data in statistical fraud models - raw and summarised data
• Extensive experience developing statistical models and scores particularly within fraud (consumer and commercial)
• Highly numerate with a relevant degree 2:1 or above
• Use of relevant analytical tooling on a day-to-day basis e.g. Python, SAS, BigQuery, Jupyter Notebook, SQL, etc.
• Previous experience of developing regression models and scores - e.g. logistic
• Day to day exposure to large/complex datasets, analytical tools, spreadsheets, analytical platforms, etc.
• Able to extract data from multiple platforms and create master datasets towards the development of statistical models
• Experience across different fraud types, e.g. consumer fraud, commercial fraud, Anti-Money Laundering, KYC.
• A team player, effective communicator with the ability to work unsupervised
• Understanding of the regulatory landscape with respect to consumer & commercial credit and the potential impact to our clients (i.e. addressing the "why" behind our decision making)

What could set you apart

• Experience of developing AI/ML based fraud solutions and models
• Day to day use of G-Suite tooling, GCP and VertexAI
• Experience of working in a credit reference agency and/or client environment e.g. Banking, Lending, Insurance, Public Sector, Utilities, etc.
• Use of LLMs to improve efficiencies in the model building process
• Use of alternative data such as open banking transactional data in statistical models
• Experience of working in a product development environment
• Experience with industry related fraud prevention tools (e.g. Hunter, Kount, SEON, ThreatMetrix, Lexisnexis, Minerva, CrossCore, TruValidate, FICO Falcon, etc.)
• Experience with standard product testing procedures (UAT, OAT, regression, etc.)
• Adoption of champion/challenger approaches and techniques
• Creative and innovative thinker
• Strong track record of delivery
• Team player with the ability to go above and beyond