Pratik GuptaHome
Back to projects

Marketplace catalog systems

Product Variation Intelligence

ML clustering + LLM validation + confidence scoring to improve product family and variation quality across categories.

ML clusteringLLM validationCatalogHITL

Problem

Noisy seller data and category-specific variation logic made scalable product grouping unreliable.

Solution

Built a multi-tier architecture: clustering, category rules, low-confidence LLM validation, and human resolution.

Outcome

Unlocked large commercial upside while balancing automation with review controls.

Architecture

A placeholder implementation path that can be expanded with screenshots, data contracts, system diagrams, and measurable results as the project matures.

01

Attribute ingestion

02

Embeddings + clustering

03

Category variation rules

04

LLM validation

05

Confidence scoring

06

Human review

Product Artifacts

Sanitized examples to demonstrate product thinking and execution style when proprietary materials cannot be shared.

  • PRD outline (problem framing, success metrics, rollout plan)
  • Workflow wireframe / journey snapshot
  • Evaluation rubric or quality checklist
  • Operational metrics dashboard mock

Metrics to Track

  • Broken grouping rate
  • Cluster precision
  • Review deflection
  • Product detail page conversion on fixed groups

Product Role

  • Owned V2 strategy
  • Secured engineering/science alignment
  • Shipped confidence-driven operational model