Machine Learning Engineer - Intelligence Group

About Smartnumbers

We are on a mission to stop fraud and improve customer authentication. Fraud is a huge problem affecting millions of people, it costs the UK nearly £7bn and represents 40% of all crime. Too often the solution has been to put in place cumbersome authentication processes that frustrate genuine customers, cause inefficiencies for organisations and fail to prevent fraud.

We are changing this by providing organisations with real‑time insight into the risk of a caller. We combine patented machine learning technology with our deep domain knowledge to prevent contact centre fraud and streamline customer experience.

We recognise that we need to work together to fight fraud, that is why we have fostered strategic partnerships with leading global organisations like BT, Genesys and Amazon. Together, we protect the UKs largest retail banks, investment banks and emergency services.

You will be part of a cross‑functional team, working across a variety of tasks from data science research and model development through to platform implementation and maintenance.

You will use your knowledge of machine learning algorithms, frameworks, and methodologies to research and develop models for our cloud‑based authentication and fraud systems, continuously iterating and evaluating model performance using appropriate metrics.

You will:

• Explore and visualise data to discover innovative features and potential data sources.
• Engineer datasets, develop data pipelines, perform feature engineering, and write code to train, deploy, monitor, and run real‑time inferences.
• Build and monitor ML models, addressing issues such as overfitting, under fitting, data leakage, and drift.
• You will use your expertise in engineering and Dev Ops/MLOps to manage our machine learning platforms using AWS Sage Maker and other AWS services.

You will:

• Design, build, and improve scalable public cloud‑based machine learning platforms.
• Develop robust data pipelines and workflows, contributing to platform reliability, scalability, and observability through effective monitoring, alerting, and performance tuning.

All our teams are given the freedom and autonomy to pick their own technology stack based on their system’s requirements and preferences. Our technology vision and strategy encourages you to try the latest innovations, and we naturally gravitate towards serverless architectures where appropriate. We value clean, maintainable and robust code for our business critical systems.

Some of the technologies currently used by the Intelligence Group are listed below - while mastery of all these areas isn’t required, familiarity with as many as possible will be advantageous.

• Infrastructure as Code:
  Amazon CDK
• ML Platform:
  Amazon Sage Maker (Sagemaker Studio IDE, Sagemaker Training / Processing / Pipeline / Endpoints, Feature Store, Model Registry, Model Monitor)
• Data Processing:
  Amazon Athena, Apache Iceberg, AWS Glue, Spark

• ML Frameworks:
  Scikit-Learn, Hugging Face
• ML Algorithms:
  Tree-based (XGBoost), Deep Learning
• Model Explainability: SHAP explanations

• Python
• Data Processing Libraries, e.g. Num Py, Pandas, Matplotlib, Librosa
• SQL

• Git Hub
• Docker
• CircleCI

• Session Initiation Protocol (SIP)
• Contact Centre as a Service (CCaaS), e.g. Amazon Connect, Genesys Cloud

Smartnumbers values diversity of experience. Candidates should have a strong combination of several of the following skills, competencies and experience:

• We expect that you will have around 2 to 3 years commercial experience across a range of platform engineering and data science responsibilities. The list below gives you an idea of the attributes you’ll need, though we’re not expecting you to have deep expertise across all aspects:
• Collaborative approach to working, preferring to discuss and brainstorm tasks with the rest of the team rather than working in isolation.
• Able to own tasks end‑to‑end, take responsibility for the quality of deliverables, and drive ML and MLOps best practices and tooling to consistently enhance our…