Lead Data Scientist

Job description

At Zego, we're on a mission to do the good thing, not the insurance thing.

Insurance hasn't changed much in over a century. The way we live, work and travel has. We're building the real-time, AI-driven infrastructure that powers innovative insurance, so good drivers get cover that works the way they actually drive.

For us, AI isn't a line on a roadmap or a buzzword on a slide. It's our operating reality, and it's how we build products that back drivers instead of the old insurance playbook.

Responsibilities

• Design and build the behavioural features and factors that drive Zego's understanding of driving quality, context and risk
• Turn raw, high-frequency GPS, accelerometer and gyroscope data into usable signal at a scale of more than 250,000 trips a day
• Research new behavioural features and detection algorithms, reading the literature, testing ideas and killing the ones that do not survive contact with real data
• Take work from hypothesis to production, prototyping in notebooks then writing the production-grade Python and SQL that scales
• Choose the simplest solution that holds, rule-based when that is enough and machine learning when it is warranted
• Lead experiments, validate impact with data, and automate insight generation
• Get features into the pricing path alongside engineers, product managers and actuaries, and translate what the sensor data shows into findings non-technical colleagues can act on

Requirements

• Quantitative foundations. An MSc, or integrated MEng or MSc, in engineering, computer science, physics, mathematics or a similar field. This is a core role and it needs strong engineering foundations
• Signal processing instinct. Working knowledge of digital signal processing or sensor physics. You can reason about noise, sampling, filtering and the physical meaning behind a signal, not just its numbers
• Production-grade delivery. A proven track record taking data science or data engineering projects into production, with strong Python and SQL (we use Snowflake) and the scientific Python stack: Polars, Pandas, NumPy, SciPy and scikit-learn. You have designed, built and maintained reliable, observable pipelines from scratch
• Machine learning judgement. Experience developing and evaluating ML models on tabular data, classification or regression, where the evaluation matters as much as the model choice
• Clear communication. You turn ambiguous problems into structured, testable ideas and share insights clearly with technical and non-technical audiences
• You work AI-first. You will use AI daily here, and we mean daily. You do not need to arrive an expert, but you do need to arrive curious, experiment fast, and take ownership of getting good quickly. People who wait to be trained will find this uncomfortable
• Experience applying signal processing to noisy, high-frequency sensor/time-series data (GPS, accelerometer, IMU) in production
• Exposure to insurance, mobile data, or behaviour modelling
• Experience with cloud platforms (AWS), containerisation (Docker, Kubernetes), or data orchestration frameworks
• Familiarity with our ML and tooling stack: MLflow for experiment tracking, DVC for data versioning, Streamlit for internal apps and demos