Senior Data Scientist - Network Simulation

Relay's network is scaling fast. The decisions that shape that growth - where to expand, how to price, where to invest - rest on models that simulate how the network behaves and how it evolves as density, geography, and operating models change. We have an MVP tool for this today; we need someone to help productionise it and build the next generation of strategic models on top of it.

You'll be a core contributor to Relay's digital twin - a network simulator that captures our unit economics end-to-end, from first-mile collection through sortation, middle-mile and last-mile. It's already in use by finance, but it's a living system: every operating-model change, new service type, or commercial scenario means upgrading a component, adding a model, or building a new view. Your job is to help keep it accurate and to extend it where the business asks new questions.

1. Evolving the digital twin. You understand each cost component deeply, you help identify where the current models are wrong or incomplete, and you upgrade them systematically.

• When a new first-mile operating model is rolled out, you upgrade that engine.

• When a new service type changes sortation cost, you understand it at a granular level and model it properly.

• You add new flows, surface new metrics, and harden the parts that matter most for the decisions being made.

2. Building new strategic models and forecasts. Volume forecasts, predictive models, new scenario tooling - work that sits inside or alongside the twin.

• You help decide what should live in the twin and what should stand alone.

• You pick the right tool for the question - classical statistical modelling, financial modelling, simulation, or ML where it earns its place.

Your primary partner is the finance team - Strategic Finance, Commercial Finance, FP&A. You translate their questions into modelling problems and give them tools that let them explore pricing, margin and projection scenarios dynamically rather than waiting for analysis to be rebuilt manually. You contribute to the roadmap - taking in their needs, helping prioritise what to build, and shipping it.

You're part of a data org of around 30 engineers, analysts, and data scientists, but embedded in the finance squad. A dedicated finance analyst within the squad partners with you directly - they build the reporting and visibility layer on top of your models and feed prioritisation back to you. You'll work alongside more senior data scientists who own the broader direction of the twin.

• You're a strong modeller who thinks in systems. You naturally break complex problems into components, understand the assumptions behind each one, and know when those assumptions need revisiting.

• You're a builder. You don't wait for requirements to be handed to you. You go and understand the problem, figure out what's needed, and ship something useful. Then you iterate.

• You're fluent in Python and SQL, and comfortable doing your own data engineering. You can pull, transform, and model data without needing someone else to prepare it for you.

• You bring strong software engineering principles into your modelling work. The digital twin is a real production system - Python, SQL, APIs, a frontend - and you'll work across that stack. You don't need to be a frontend specialist, but you should be at ease in a production codebase: writing tested, readable code, designing modular components, shipping changes that someone else can pick up next week.

• You have experience with financial modelling, forecasting, or simulation, ideally in a context where your models informed commercial or strategic decisions - not just produced analysis. You know your way around ML and use it where it materially improves accuracy.

• You communicate clearly with non-technical partners. Finance will rely on your models to make decisions. You need to explain what they do, where they're reliable, and where they're not.

• You care about the problem more than the technique. You'll use whatever modelling approach fits. The point is accuracy, speed, and usefulness - not methodological elegance.

• You take ownership of your work. You manage trade-offs and hold yourself accountable for whether the models are actually driving better decisions.

• Logistics or delivery network experience is a plus, but what matters more is the ability to learn a complex operational domain quickly and model it well.