Overview
A multi-part intelligent support system combining machine learning, computer vision, retrieval-augmented generation, and agentic workflows.
The system combines return-risk prediction, product-image classification, and a LangGraph-based support agent into a single workflow. The support agent can ground policy questions through a local FAISS knowledge base and invoke the trained ML models as real tools.
The project focuses on connecting individual machine learning models into a practical, tool-using AI system rather than treating them as isolated experiments.
What this project explores
Agentic AI
RAG
LangGraph
Machine Learning
Computer Vision
E-commerce