Data Scientist - Detections

Who are we?

Hello! 👋 We are Ravelin! We are a company specializing in fraud detection, using advanced machine learning and network analysis technology to tackle big problems. Our objective is to secure online transactions and assure our clients can reliably serve their customers.

We enjoy what we do! We are a friendly team committed to our culture and our values of empathy, ambition, unity, and integrity. We highly cherish work/life balance and we adhere to a flat hierarchy within our company. Join us, and you will rapidly learn about cutting-edge technology and work with some of the most brilliant and pleasant individuals around -

If this is your cup of tea, we would be thrilled to hear from you! For further details, visit our and see if you are interested in helping us prevent crime and protect the world's largest online businesses.

The Team

You will be a part of the Detection team. This team is tasked with maintaining low fraud rates to keep clients satisfied by continually training and implementing machine learning models. Our goal is to ensure model deployments are as simple and error-free as code deployments. Google’s serves as our guide.

Our models are trained to identify various types of fraud, using multiple data sources and techniques in real time. The prediction pipelines strictly uphold Service Level Agreements, every prediction must be returned in under 300ms. When models are not working as expected, it is the Detection team's task to investigate the reasons.

The Detection team is the backbone of Ravelin’s success. They closely collaborate with the Data Engineering Team that constructs infrastructure and the Intelligence & Investigations Team that liaise with clients.

The Role

We are currently looking for a Data Scientist to aid in training, deploying, diagnosing, and evaluating our fraud detection models. Our ideal candidate is pragmatic, accessible, and rich in experience and knowledge rooted in previous failures.

Evaluating fraud models is challenging; often, we do not get labels for 3 months. You will need to exercise your judgement when examining ambiguous fraud cases and when probing the truthfulness of the model itself.

We must build sturdy models that are capable of updating their beliefs when they encounter new fraud methods: our clients expect us to stay one step ahead of fraud, not lagging. You will be equipped with the tools, space, and guidance you require to build world-class fraud detection models.

The work is not all about novel research. Day-to-day work involves making safe, gradual progress towards better models for our clients. The ideal candidate is eager to engage in both aspects of the job – and understand why both are necessary.

Responsibilities

• Expand our model evaluation and training infrastructure.
• Create and implement new models to detect fraud while upholding SLAs
• Contribute new features to our production infrastructure
• Research innovative tactics to disrupt fraudulent behavior
• Analyze model performance issues (utilizing your expertise in debugging models).

Requirements

• Experience in developing and implementing ML models using the Python data stack (numpy, pandas, sklearn).
• Strong analytical skills.
• Ability to collaboratively work with colleagues outside your team, such as client support teams or engineering.
• Aptitude in communicating complex technical ideas to various audiences.
• The capacity to manage your workload and prioritize accordingly.
• Comfortability in working with a hybrid team

Desirables

• Understanding of software engineering best practices (version control, unit tests, code reviews, continuous integration/continuous delivery) and their application to machine learning engineering.
• Experience with Tensorflow and deep learning
• Experience with Kubernetes and ML production infrastructure.
• Experience with Go, C++, Java or another systems language.

Benefits

• Flexible working hours, hybrid working model, office located in Old Street stocked with snacks and drinks and a £500 home office budget
• Company shares options
• 25 days holiday + bank holidays + an additional day off for every year of service (up to 5) + 1 extra day off for cultural reasons
• Extra monthly company-wide days off - the Wellbeing & Learning Days
• £1000 annual wellbeing budget to spend through Heka
• Mental health support via Spill
• Comprehensive medical cover with AXA which includes pre-existing conditions
• Pension Scheme with Aviva
• Enhanced parental benefits
• Company outings, team socials, and a budget for impromptu events that anyone can organize
• Ravelin Gives Back (RGB) - monthly charitable donations and regular volunteering opportunities
• Bi-weekly team lunches with a randomized group of people from across the company, either virtually (via Deliveroo) or in-person
• Access to BorrowMyDoggy
• Support for purchasing bicycles tax-efficiently through the Cycle-to-Work scheme
• Weekly board game nights

Job offers may be rescinded if candidates do not pass our pre-employment checks: unspent criminal convictions, employment verification, and right to work.