ML Engineer — Demand Forecasting & Consumer Insights
About the Role
We are building an AI-native product creation pipeline for a global apparel and footwear company — a 90-day sprint to design, manufacture, and place real products on retail shelves without using the traditional 18-month fashion calendar. The proof of concept is simple: 5,000 units, one category, 90% full-price sell-through.
The ML Engineer owns the intelligence layer that makes this possible. You will build the models that tell the team what consumers want before they know it themselves — and the forecasting logic that determines exactly how many units to produce, in which sizes, for which markets. Get this right and the pilot sells through cleanly. Get it wrong and the warehouse fills up with unsold backpacks.
What You Will Do
• Build or configure consumer insights models that synthesize real-time social media signals, search trends, and behavioral data into actionable design briefs — replacing the 9-month, $200K research agency process with near-real-time intelligence
• Develop demand forecasting models at the SKU, size, and geography level to inform micro-batch production decisions — starting with 5,000 units for the pilot and designed to scale
• Integrate data inputs from social platforms (TikTok, Instagram, Google Trends) and commercial trend tools (Heuritech, WGSN, Stylumia) into a unified signal layer
• Work with the AI Product Manager to define the data architecture for the pilot — lightweight enough to move fast in Phase 1, structured enough to scale in Phase 2
• Validate model outputs against real sell-through data as the pilot progresses and iterate accordingly
• Document model logic and data pipelines so the approach is repeatable across categories and brands
What You Need
• Production-level experience building demand forecasting or consumer demand sensing models — time-series forecasting, regression, or ML-based approaches; you have shipped models that drove real inventory or production decisions
• Hands-on experience with retail or consumer data — SKU-level sales data, social listening data, search trend data, or equivalent; you understand how noisy and inconsistent this data is and how to work with it anyway
• Familiarity with at least one commercial forecasting or trend intelligence platform — o9 Solutions, Blue Yonder, Heuritech, Stylumia, or equivalent
• Strong Python skills and comfort working in cloud environments — models need to run in production and produce outputs the team can act on
• Ability to communicate model outputs in plain language to non-technical team members — the tiger team includes a designer and a marketing lead; you need to translate forecast confidence intervals into decisions they can make
What You Do Not Need
• Experience in fashion or apparel — consumer behavior data is consumer behavior data; retail, CPG, or e-commerce forecasting experience is equally relevant
• Deep SAP or ERP integration experience — the pilot will use manual file transfers where needed; backend integration is Phase 2
• A large team or large compute budget — the pilot is lean by design; the models need to work with limited data and limited infrastructure
Nice to Have
• Experience with social commerce data pipelines — TikTok and Instagram engagement signals as demand proxies
• Familiarity with micro-batch production economics — unit economics of 50–500 unit test runs versus full production scaling
• Experience building consumer persona or segmentation models using LLMs
Location & Commitment
Primarily remote. Periodic travel to Costa Mesa, CA for tiger team sprints. Full-time commitment for the duration of the pilot with expectation to continue into Phase 2 if the pilot succeeds.
Never miss an opportunity! Create an alert based on the job you applied for.
