AI Demand-Sensing and Waste Reduction Leader

The mandate

The leader integrating an acquired forecasting team resigned after holiday and weather-driven demand shifts produced simultaneous stockouts and spoilage across fresh categories. Store planners overrode recommendations inconsistently, but model teams excluded those interventions from their accuracy story, leaving no owner for the combined economics of availability, markdown, waste and trust.

The interim must take the Bangkok seat within three weeks and hold it for fifteen months. A permanent search begins after the first full seasonal cycle under rebuilt controls, expected around month nine, with six weeks reserved for the successor to chair a category review and approve the next regional replenishment release.

Handover requires priority fresh categories to remain inside signed availability and waste corridors for two cycles, forecast and override lineage to be reproducible, shelf-life and substitution constraints to operate in ordering, two disruption exercises to pass, and the permanent appointee to accept the category-region exception register. Average forecast accuracy alone will not close the assignment.

The interim may stop model or category deployment, set order-policy boundaries, redesign planner roles, quarantine unreliable features, appoint temporary leads and move up to THB 900 million within the sanctioned recovery. Supplier delisting, permanent executive hiring, pricing policy, network capital and spend above THB 250 million per decision require council approval; food-safety controls cannot be overridden.

Store-format strategy, loyalty replacement, supplier contracting and warehouse automation are explicitly out of scope. The brief covers demand sensing, replenishment recommendations, fresh constraints, planner decisions, outcome measurement and the release-and-governance system needed for reliable scale.

Why this seat is open

The failed seasonal cycle exposed a gap between statistical ownership and commercial-operating consequence. The integration leader’s exit left acquired scientists and retail planners defending different success measures while waste continued. Temporary authority is required to rebuild one decision system before permanent leadership governs the next category expansion.

What you will own

• Reconstruct failures by store, category, shelf life, promotion, weather, substitution, supplier constraint, model version and planner intervention.
• Set category-region decision objectives balancing availability, waste, markdown, margin, service and food-safety constraints.
• Decide where recommendations automate, require planner approval, operate inside bounds or remain excluded because evidence is sparse.
• Rebuild features and evaluation around season, local events, weather, lead time, freshness, substitution and censored lost-sales demand.
• Establish override governance that captures reason, timing, quantity, outcome and feedback without penalising justified human challenge.
• Command disruption exercises covering supplier shortfall, heat event, promotion error and distribution delay before broader rollout.
• Transfer the model registry, category playbooks, economics, override evidence and release calendar through a successor-led cycle.

Candidate qualifications

• Held senior authority for demand forecasting or replenishment AI across fresh, grocery, pharmacy or another perishable retail network.
• Improved availability and waste together, with realised category economics rather than forecast accuracy as the primary evidence.
• Integrated an acquired or external data-science team with planners and store operations during a failed or contested rollout.
• Modelled shelf life, substitution, promotion, weather and constrained supply in production ordering decisions.
• Designed human override capture that improved learning while preserving accountable planner intervention.
• Handed a recovered demand platform to permanent leadership after a full seasonal cycle and disruption exercises.

Non-negotiables

• Can start in Bangkok within three weeks, work on site initially and travel monthly to stores, markets or distribution centres.
• Will preserve planner dissent and stop category automation when waste, availability or food-safety corridors fail.
• Brings direct perishable decision experience; generic retail analytics or non-operating forecast research is insufficient.
• Must disclose current retailer, supplier, forecasting-vendor and data-provider relationships before performance evidence is shared.

1. 49 words maximum. State your earliest Bangkok start and any commitment incompatible with a fifteen-month executive assignment.
2. 49 words maximum. Describe one fresh-category model that improved accuracy while worsening waste or availability and what you changed.
3. 49 words maximum. Which planner-override pattern would make you pause automated replenishment?