AI-Powered Product Recommendation System

Overview

Architected an AI-powered recommendation system that analyzes millions of Amazon product reviews to help users quickly discover the most relevant and high-quality items through semantic search and LLM-based understanding.

Key Technologies

• Data Processing: PySpark, BigQuery
• Machine Learning: Text-embedding-005, ScaNN, Google Gemini, LangChain
• Infrastructure: GCP (Cloud Run, BigQuery, Cloud Storage), Terraform, FastAPI

Achievements

• Developed a PySpark ETL pipeline to clean, tokenize, and embed reviews (768-dim via text-embedding-005)
• Designed hybrid retrieval engine improving nDCG@3 by +21% (0.85 vs 0.70) with MRR = 0.88
• Integrated RAG-based sentiment analysis achieving 88% accuracy and 4.3/5 relevance score
• Sustained ~6s query latency with 92% product-category coverage across 500 test queries

Technical Details

• Hybrid retrieval using ScaNN with approximate nearest neighbors (TreeAH + AVQ)
• Metadata filtering and reranking via FastAPI microservice
• Scalable infrastructure provisioned with Terraform on GCP
• Explainable recommendations through LLM-based feature summarization