Director of Decision Science

Stord is The Consumer Experience Company, powering seamless checkout through delivery for today's leading brands. Stord is rapidly growing and is on track to double our revenue in the next 18 months. To meet and exceed this target, Stord is strategically scaling teams across the entire company, and seeking energetic experts to help us achieve our mission.

By combining comprehensive commerce-enablement technology with high-volume fulfillment services, Stord provides brands a platform to compete with retail giants. Stord manages over $10 billion of commerce annually through its fulfillment, warehousing, transportation, and operator-built software suite including OMS, Pre- and Post-Purchase, and WMS platforms. Stord is leveling the playing field for all brands to deliver the best consumer experience at scale.

With Stord, brands can increase cart conversion, improve unit economics, and drive sustained customer loyalty. Stord’s end-to-end commerce solutions combine best-in-class omnichannel fulfillment and shipping with leading technology to ensure fast shipping, reliable delivery promises, easy access to more channels, and improved margins on every order.

Hundreds of leading DTC and B2B companies like AG1, True Classic, Native, Seed Health, quip, goodr, Sundays for Dogs, and more trust Stord to deliver industry-leading consumer experiences on every order. Stord is headquartered in Atlanta with facilities across the United States, Canada, and Europe. Stord is backed by top-tier investors including Kleiner Perkins, Franklin Templeton, Founders Fund, Strike Capital, Baillie Gifford, and Salesforce Ventures.

This is Stord's first dedicated Decision Science leadership role, a hands-on Data Science Director role for someone who wants to build a full-stack analytics and machine learning function from the ground up. Stord processes $10B+ in commerce annually across fulfillment, warehousing, and software platforms, generating rich datasets spanning consumer behavior, warehouse operations, and parcel networks.

You'll turn that data into a competitive advantage: owning the path from raw data to production model to adopted business decisions, not just delivering dashboards or one-off analyses.

You'll build on Stord's modern data stack (GCP, BigQuery, dbt) and help define how the Decision Science function adopts Claude and agentic AI as core infrastructure, not just a productivity add-on, but part of how the team scales analytics, experimentation, and machine learning across the business.

How We Work

Stord runs on Google Cloud Platform, and Claude (Anthropic) is our primary AI platform, company-wide. This function won't just use AI tools, you'll help define how a modern Decision Science team works alongside agentic AI: governed metrics and semantic layers that agents query first, documented "skills" for safe self-service analytics, and clear human ownership of metric definitions and high-stakes model review.

If you want to build a data organization from the ground up on a genuinely AI-forward stack, this is a rare opportunity to set that foundation.

Key Responsibilities:

• Machine Learning Portfolio : Design and productionize models in delivery prediction, carrier routing, demand forecasting, exception management, and churn analytics, owning the full model lifecycle from notebook to production .
• Experimentation Platform : Build a self-serve A/B testing and causal inference platform the broader business can use independently, not a one-off analysis service.
• Advanced Analytics: Conduct segmentation, behavioral analysis, and cohort analysis supporting product and operations decisions.
• AI-Augmented Decision Science: Partner with the Head of AI to integrate model outputs into AI-native products, and shape how the team uses agentic AI (governed semantic layers, agent-based querying, skill/prompt documentation) to automate routine analysis and free the team to focus on forecasting, causal inference, and ML.
• ML Adoption: Ensure models drive actual business decisions, translating outputs into actionable workflows, not dashboards.
• Team Leadership: Hire, develop, and lead a high-performing Decision Science team as a player-coach: hands-on individually while building the team around you.
• Year 1 Success Criteria
• Team staffed and contributing
• Five+ production models with quantified business outcomes
• Live, adopted experimentation platform
• Business stakeholders actively using model outputs
• Full commerce data stack (consumer, fulfillment, parcel) actively modeled
• Year 2 roadmap defined with organizational buy-in
• Required Qualifications: Technical Depth
• Practitioner-level machine learning: design, build, and evaluate models independently, end to end (notebook to production)
• Expertise in supervised learning, time-series forecasting, segmentation, recommendation systems, and lift measurement
• Strong experimentation design and causal inference skills, with the statistical fluency to communicate results to non-technical stakeholders
• Proficiency with GCP and BigQuery (or a comparable cloud data warehouse); hands-on experience with dbt, semantic layers, and modern ML tooling (BigQuery ML, Vertex AI, or equivalent)
• Familiarity operating in or building an LLM-augmented analytics environment, e.g., semantic layers, agent-based querying, prompt/skill documentation, or comparable AI-assisted self-service analytics tooling (Claude, Anthropic, or similar)
• Leadership
• Player-coach mindset: hands-on in a small team environment, with a track record of developing junior talent
• Ability to establish cross-functional credibility and drive ML/AI adoption in skeptical or immature data environments
• Fluency translating technical work into business language, lift, cost per unit, margin, retention - not statistical jargon
• Business Instinct
• Clear understanding of how Decision Science connects to revenue and cost

Preferred Qualifications:

• Experience operating at the intersection of physical operations and software (supply chain, logistics, 3PL, fulfillment, retail ops, or similar "unglamorous," real-world data environments)
• Comfort with messy, real-world operational data rather than a mature, pre-cleaned platform
• Track record driving adoption of models and experimentation in complex, multi-stakeholder organizations
• Experience going from individual contributor to Director, or building a data science function from zero to one