←All projects/Freelance · Jul 2026 – Aug 2026 · E-commerce data engineering
Real project
Eight product catalogs merged into one online store.
A real, anonymized freelance engagement that transformed 28,646 source product records into reviewed, validated and WooCommerce-ready catalog workflows.
28.6Kproduct records processed
The problem
A client needed furniture and home products from eight websites in one WooCommerce store. Each source structured products, variants and images differently.
What I built
A resumable pipeline that scrapes, standardizes, checks and batches products for import into the client's Hostinger-hosted store, with review queues for unclear cases.
My role
Built the whole workflow end to end as a freelancer.
Result
28,646 product records processed. 5,342 media references reconciled with zero recorded failures.
Why it was hard
Eight different schemas and site structures
Large image volumes and unreliable downloads
Imports had to be rerunnable without duplicate SKUs
The client needed to consolidate furniture and home-product catalogs from multiple websites into a single e-commerce store. Product structures, categories, variants, prices, descriptions and images differed by source, while the final imports had to be complete, reviewable and safe to rerun.
MY ROLE
Where I created leverage.
I built the workflow end to end: source-specific extraction, checkpoints, canonical product modeling, category reconciliation, image preparation and upload, WooCommerce import generation, batching and audit reports.
SANITIZED WORK SAMPLE
Different catalogs converge on one import contract.
The example uses generic product language while preserving the real transformation and validation pattern.
01 / SOURCE RECORDTITLE + VARIANT + REMOTE IMAGE
Eight catalogs arrived with different structures and completeness.
02 / CANONICAL PRODUCTSKU + CATEGORY + ATTRIBUTES
A shared product model makes mapping and review consistent.
03 / IMPORT EVIDENCEBATCH + MANIFEST + AUDIT TOTAL
Every output can be reconciled before WooCommerce import.
+Generic example only. Client identity, store data, source retailers and product-level records are withheld.
Technical details
SYSTEM FLOW
From friction to a repeatable flow.
Catalog discovery
Source adapters
Checkpointed extraction
Canonical model
Category mapping
Media reconciliation
WooCommerce batches
Audit summary
CONSTRAINTS
The difficult parts.
Eight source catalogs with different schemas and navigation patterns
Large image volumes and unreliable remote media responses
Client-specific category rules and products requiring human review
Imports needed to be resumable, auditable and safe against duplicate SKUs
DESIGN DECISIONS
How I approached them.
Separated source-specific collectors from one canonical product schema so new catalogs could be added without redesigning the migration layer.
Used checkpoints and manifests to preserve completed work and make long-running extraction and media operations resumable.
Created review queues for ambiguous categories, exclusions and missing assets instead of silently forcing uncertain products into the store.
Validated reconciliation totals, SKU uniqueness, allowed categories, images and batch readiness before marking an output safe to import.
OUTCOMES
Project outcomes
28,646
source product records
Verified across eight extracted catalog datasets.
26,832
catalog rows audited
Checked for duplication, completeness, validity and category quality. The other 1,814 source records were products the client chose not to publish.
5,342
media references reconciled
Tracked across completed WordPress media workflows with zero recorded failures.