Sr. Product Manager Tech, Product Quality, Perfect Order Experience (POE)

Amazon customers expect every product they receive to be authentic, undamaged, and exactly as described. The Global Product Verification (GPV) team builds AI-powered computer vision and machine learning systems that detect product quality defects - counterfeit, expired, damaged, and wrong-item products - before they reach customers, protecting trust at fulfillment center scale.

As a Senior Product Manager-Tech for GPV, you will own the product strategy and technical architecture for Amazon"s Unified Image Pipeline - a multi-modal CV/ML system that fuses RGB imaging, X-ray analysis, OCR, and visual similarity with confidence-based routing to automatically detect and enforce product defects across the fulfillment network. You will drive end-to-end ownership of the product quality investigation journey, making architectural trade-off decisions across ML model design, hardware-software integration, and pipeline orchestration. You will partner with applied scientists, software engineers, hardware teams, and operations stakeholders to define the technical vision, prioritize capabilities, and ship solutions that reduce product condition complaints for hundreds of millions of customers worldwide. This role offers the opportunity to shape how Amazon verifies product authenticity and condition at global scale through AI/ML systems.

Key job responsibilities

1. Own the technical product vision and multi-year roadmap for GPV"s CV/ML defect detection portfolio, including the Unified Image Pipeline, mobile investigation tools, authentication systems, and product quality trust signaling

2. Make architecture-level trade-off decisions across ML model design (CNN embeddings, multi-modal fusion, confidence routing), imaging modalities (RGB, X-ray, OCR), and hardware-software co-development

3. Define confidence-based enforcement frameworks that translate ML model outputs into automated product decisions at scale (auto-enforce, human-in-the-loop, log-only thresholds)

4. Drive buy vs. build decisions for authentication technology, evaluating external licensing against proprietary ML + hardware approaches based on capability assessment and data risk analysis

5. Own input and output metrics (TPR, AHT, auto-resolution rates, complaint reduction, GMS unlock) and build data-to-product-decision architectures that link model performance to customer outcomes

6. Collaborate with applied scientists, SDEs, hardware engineers, operations, legal, and business stakeholders to deliver solutions that reduce 347M+ annual product condition complaints

7. Develop product plans with clear measurable success criteria, phased rollout strategies, and mechanisms to communicate progress against the roadmap to senior leadership