Senior Data Scientist, Planning and Forecasting

THE ROLE

Senior Data Scientist, Planning and Forecasting

Quince is building its own supply chain planning science capability from scratch. This involves demand forecasting at multiple geographic scales, methodology-agnostic forecasting tournaments, inventory placement optimization across a growing international network, vendor performance modelling, raw material signal generation. The agenda is wide, the data is rich, and the business consequence is direct.

As a senior data scientist on this team, you’ll own modelling workstreams within that agenda. You’ll work alongside the charter leadership setting the science roadmap and the engineering team building the platform around your work. This role comes with ample scope to do real, end-to-end science, and enough engineering support that models you build actually run.

We expect AI-native science. Modern forecasting and OR work makes regular use of LLM-aided EDA, agentic feature engineering, and AI-augmented experimentation infrastructure; we want a scientist who already builds this way.

The ideal candidate is a senior data scientist with 5-8 years of production experience, who has owned modelling workstreams from problem framing through deployment and iteration. They are opinionated about methodology, what it takes to design a defensible forecasting tournament, and how to handle demand patterns that don’t behave.

They identify as a scientist but they build like an engineer when they need to. They can write the feature pipeline, run the experiment, and ship the model serving without waiting for a counterpart to translate. They earn the trust of business stakeholders through the clarity of their work, not the polish of their slides.

They are AI-native in their science workflow, including LLM-aided exploratory analysis, agentic feature discovery, and AI-assisted experiment design, with the rigor to validate AI suggestions before they become decisions. They are at home in a tournament framework that includes statistical, ML, and AI-driven models on equal footing.

Responsibilities

• Modelling Workstreams
  • Own a meaningful slice of the science roadmap; for example, demand forecasting for a product category, vendor performance modelling, inventory placement optimization, or raw material signal generation.
  • Run rigorous, well-documented experiments: data prep, feature engineering, model selection, evaluation, and the discipline of knowing when a result is real and when it isn’t
  • Bring methodological breadth to bear: classical, ML, and AI-driven approaches evaluated on their merits within a tournament framework, not as advocacy for any one school
• AI-Native Science
  • Use LLM-aided EDA, agentic feature discovery, and AI-augmented experiment design as default tools, applying high judgment to validate AI suggestions before they shape decisions.
  • Contribute to the team’s standards for evaluating AI-driven forecasting and OR models alongside classical and ML approaches
• Production Discipline
  • Ship models into production, not into reports. Own the deployment, the monitoring, and the iteration loop that keeps a model accurate as the world shifts under it
  • Partner with the Planning Tools engineering team on what the platform needs to expose for your work (feature stores, evaluation harnesses, model registry) and what your work needs to be designed for in turn
• Business Partnership
  • Work directly with planning operators on the problems they’re trying to solve; translate their operational reality into well-defined modelling problems, and translate your model outputs into decisions they can act on
  • Hold honest opinions about what science can and can’t solve, and surface them clearly enough to influence the roadmap

Qualifications

Required:

• 5-8 years of applied data science or operations research experience, with a demonstrable track record of shipping models that moved business metrics
• Real depth in forecasting methodology (statistical, ML, and AI-driven) with informed opinions about when to reach for which
• Substantive experience with at least one OR domain: inventory optimization, vendor lead-time modelling, transportation mode selection, or equivalent
• Engineering fluency to own your work end-to-end: feature pipelines, experiments, model serving
• AI-native science practice you can speak to specifically (LLM-aided EDA, agentic feature discovery, AI-augmented experimentation) with examples of when you trusted AI output and when you overrode it

Preferred:

• Advanced degree in a quantitative field (Statistics, CS, Operations Research, Engineering, Economics) preferred