Product Content Generation / E-commerce AI
Product data into publishable copy.
An AI pipeline for enriched catalog data, product descriptions, and category-aware marketing content.
Custom parsers collect catalog data and category-specific product attributes.
Illustration of the content-generation pipeline. Example product content is illustrative.
- Python
- NestJS
- FastAPI
- LangChain
- Modal
- AWS Lambda
- DynamoDB
How I approached the work
Engineering decisions.
Constrain generation with product context
- Problem
- Bulk product copy needs consistent category context and quality checks across catalog entries.
- Decision
- Combined product-data enrichment with category-conditioned prompt chains, semantic deduplication, keyword scoring, and classification.
- Result
- A serverless pipeline for product titles, descriptions, and marketing content, deployed on Modal.
The work.
I built an AI content pipeline for e-commerce catalogs. It combines product-data extraction with LLM generation, category signals, and quality checks to support bulk creation of titles, descriptions, and marketing content.
My contribution
Catalog enrichment
Developed scraping infrastructure and custom parsers to collect and enrich product information.
Controlled generation
Designed prompt chains and conditioned templates using category-specific signals and keyword optimization.
Quality at scale
Added semantic deduplication, keyword scoring, and classification, and deployed serverless processing on Modal.