Director of Decision Science
The Opportunity
Stord is the commerce enablement platform that powers $10B+ in commerce annually for some of the world's leading brands. We sit at the intersection of physical operations and software - running fulfillment centers, parcel networks, and the technology stack that ties it all together. Few companies have data like this. On the consumer side, we see the full pre and post-purchase journey: browse and cart behavior, order placement, fulfillment events, delivery outcomes, returns, and repurchase. Inside the warehouse, we capture every pick, pack, and ship event across our fulfillment network - throughput, accuracy, labor efficiency, exception rates. Across our parcel network, we see carrier performance, delivery prediction, SLA adherence, and cost at the shipment level. This is not a single domain dataset. It is the full commerce stack, end to end.
Decision Science is the function that turns that signal into competitive advantage. The modeling opportunities here are genuinely rich: delivery prediction, carrier routing optimization, demand and volume forecasting, brand-level churn and performance analytics, exception management, personalization. The opportunity is to build a function that develops models the business trusts, adopts, and acts on - and that makes Stord smarter with every order we process.
What You'll Own
ML model portfolio - Design, develop, and productionize ML models that drive measurable operational outcomes. Priority domains include delivery prediction (EDD), carrier routing optimization, demand and volume forecasting, exception management, and brand-level churn and performance analytics.
Experimentation framework - Build and own Stord's experimentation capability. That means rigorous A/B test design, lift measurement, causal inference where appropriate, and a framework the rest of the business can use to run experiments without coming to your team for every one.
Advanced analytics and segmentation - Own the analytical depth that supports product, operations, and commercial decisions - customer and brand segmentation, behavioral analytics, cohort analysis.
ML adoption - Ensure models are actually used. This means translating outputs into language and workflows the business acts on, not publishing results to a dashboard no one reads.
Team - Build and lead a high-performing Decision Science function. Hire well, develop the people you have, and create an environment where strong data scientists do their best work.
AI partnership - Work alongside the Head of AI to ensure ML model outputs are accessible to AI-native products and that the Head of AI's roadmap has the model-driven signal it needs to be effective.
What Success Looks Like in Year 1
The team is staffed and operating well - data scientists are hired, onboarded, and contributing at pace.
A portfolio of ML models is in production - we are targeting five or more models running in live operational or commercial contexts, each with a quantified business outcome: cost reduction, accuracy improvement, a routing decision that changed, a churn signal that was acted on.
An experimentation framework is live and adopted - Operations and Product teams can run and interpret A/B tests without routing every experiment through your team.
Business stakeholders across Operations and the Commerce product group are actively using model outputs in their decisions - this is not a nice-to-have, it is a success criterion.
The full commerce data stack - consumer, fulfillment, and parcel - is being actively modeled, not just the most obvious domain.
The Decision Science roadmap for Year 2 is defined, credible, and has organizational buy-in.
What We Are Looking For
You are a player-coach. You have the depth to design and build models yourself and the leadership instinct to grow a team that does it without you. You are not an ivory tower data scientist and you are not a pure people manager. You are the person who can sit with an operations leader, understand a business problem, translate it into a modeling opportunity, build it, and then make sure it actually changes how decisions are made.
Technical Depth - Practitioner-level ML - you can design, build, and evaluate models yourself, not just manage people who do. Supervised learning, time-series, segmentation, recommendation systems, and lift measurement are all in your toolkit. Experimentation methodology - you know how to design a proper experiment, size it correctly, account for confounders, and communicate the result in plain English. P-values are not your primary currency. Full model lifecycle - you have taken models from raw data to something running reliably in a production environment. You understand the gap between a notebook result and a model people depend on. Modern data platforms - comfortable working with BigQuery or equivalent cloud warehouses, familiar with dbt or semantic layer concepts, not dependent on a perfect data engineering handoff before you can start building.
Leadership and Team Player-coach commitment - willingness to be hands-on is non-negotiable. This is a small team. You cannot manage from a distance. Develops junior talent - you can take a capable data scientist and make them better. You know what good looks like and how to close the gap. Cross-functional credibility - you build trust with operations leaders, product managers, and engineers who are not data people. They need to believe in your models before they will change how they work. Commercial and Business Instinct - Business-language first - you frame model value in outcomes the business cares about, not statistical metrics. Lift, cost per unit, margin improvement, retention. Not precision-recall curves. Connected to the commercial layer - you understand how Decision Science connects to revenue and cost, not just analytics. You can make the case for your team's roadmap in a budget conversation.