Applied AI

Agentic RAG System for Product Discovery

Designed and prototyped an agentic RAG-based AI system to improve product discovery and enable richer customer interactions. The system uses MongoDB Vector Search for semantic retrieval, MLflow for experiment tracking and model management, and Streamlit for interactive prototyping. The architecture supports multi-step reasoning, tool use, and grounded generation for complex product queries.

Domain

Retail / E-Commerce

Methods

  • Retrieval-Augmented Generation
  • Agentic Workflows
  • Vector Search
  • Prompt Orchestration
  • LLM Evaluation

Tools

  • Python
  • MongoDB Vector Search
  • MLflow
  • Streamlit
  • LangChain

Impact / Outcome

Enabled richer product discovery and customer interactions through agentic AI with grounded generation.