LLM Application Development Intern (Java Backend & RAG) at Allied Medical Limited

Developed a Java/Spring Boot backend for an AI-assisted parts identification workflow using Spring AI, Azure OpenAI GPT-4o, PostgreSQL/pgvector, MongoDB, Docker, and a React front API. Extended a multimodal RAG pipeline across 50,000+ ERP/stock-code records, product manuals, PDFs (text + visuals), reference images, and customer attachments using structured retrieval and vector search. Built automated ingestion and embedding lifecycle management by monitoring OneDrive-synced folders with archive/version control, source-keyed refresh, and stale vector removal. Produced structured LLM outputs for anchor product identification, candidate part recommendations, attachment analysis, confidence ranking, and follow-up guidance to support human-in-the-loop review. Integrated an event-driven front workflow using tags, webhooks, asynchronous processing, and REST APIs to run multimodal RAG analysis and post editable internal comments while keeping customer replies human-reviewed.