Data Scientist Forecasting Jobs in California (NOW HIRING)

THE ROLE

Staff Data Scientist, Planning and Forecasting

Quince is building its own supply chain planning science capability from scratch. This includes demand forecasting at multiple geographic scales, methodology-agnostic forecasting tournaments, inventory placement optimization across a growing international network, vendor performance modelling, and the raw material signal generation that links the forecast back to procurement before failures happen.

The Staff Data Scientist sets the science charter and writes the roadmap, driving its load-bearing components end-to-end. You'll be the deep specialist on a broad mandate (the forecasting tournament implementation, the inventory placement model, the vendor performance system) with full ownership of the methodology, the production model, and the iteration loop.

You'll work closely with charter leadership, mentor the DS3s and DS2s on the team, and partner directly with planning operators.

We expect AI-native science. The methodology you bring should already include LLM-aided exploratory work, agentic feature engineering, and AI-augmented experimentation. Your standards for what counts as a real result should be high enough that AI assistance accelerates rather than dilutes them.
The ideal candidate has roughly a decade of applied data science or operations research experience, with deep expertise in one or two domains relevant to supply chain planning. They've owned modelling workstreams end-to-end across multiple companies or products, and they have the craft and the patience to take a hard problem and stay with it until the model actually moves the metric.

They are excellent at being given an ambiguous problem and solving it exceptionally well. They mentor junior scientists; they earn trust with operators; they argue for the right methodology even when it's the harder one to implement.

They are AI-native in their science workflow as a matter of course. They use LLMs in EDA and feature work, run agentic loops where they make sense, evaluate AI-driven models on equal footing in a tournament framework, and have the rigor to keep AI assistance from quietly degrading the science.

Responsibilities

Workstream Ownership

• Own the science workstreams end-to-end: the forecasting tournament implementation, the inventory placement optimization, the vendor performance system, or the raw material signal pipeline - across methodology, production model, and iteration loop
• Hold the methodological standard for your area: when to use which model class, what constitutes a defensible evaluation, what to do when the data is too sparse or too noisy
• Partner with the Planning Tools engineering team on what your workstream needs from the platform, and on the constraints production places back on what you can build

Methodology & Rigor

• Bring depth across statistical, ML, and AI-driven methods; evaluate them on their merits within a tournament framework rather than advocating any one school
• Set the standard for experimentation discipline within the science team: clean splits, honest backtests, the willingness to reject your own hypothesis
• Drive AI-native science workflow (LLM-aided EDA, agentic feature discovery, AI-augmented experiment design) with rigor to match

Mentorship

• Mentor scientists within the team; raise the methodological floor of the people around you through code review, design discussion, and direct teaching

Business Partnership

• Partner with planning operators on the problems within your workstream; translate their operational reality into well-defined modelling problems, and your model outputs into decisions they can act on
• Hold the methodological line in business conversations: educate operators on what your models can and can't support, and push back on misclassified signals or over-fitted requests

Qualifications

Required:

• 8+ years of applied data science or operations research experience, with deep expertise in one or two domains relevant to supply chain planning
• Demonstrated ownership of modelling workstreams end-to-end; from problem framing through production deployment, iteration, and measured business impact
• Real methodological breadth across forecasting and OR: classical, ML, and AI-driven approaches, with informed opinions about which to reach for
• Engineering fluency to own your work end-to-end: feature pipelines, experiments, model serving
• AI-native science practice you can speak to in detail with examples of where AI tooling materially changed how you do science, and where you held your standards against it
• Track record of mentorship: scientists who became better because they worked with you

Preferred: 

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