"Prince AMD Build team on AMD AI League Hackathon"
Team Idea
I am building an AI-powered career transition assistant that helps students and professionals figure out the right skills to switch into a target role. Built with Python, LangGraph, Neo4j, and Google Gemini 2.5 — same stack as my Roleneo project where I run 6 parallel specialist agents (pathfinder, market researcher, social scout, etc.) over a GraphRAG layer on O*NET occupational data. I'm porting the inference + retrieval pipeline to AMD Developer Cloud to test how the multi-agent system performs on AMD MI GPUs with ROCm vs running on Google APIs. Goal is to benchmark whether local AMD GPU inference can handle vector search + Cypher graph queries + LLM calls simultaneously without lag. This is my first time deploying a full agentic AI stack on AMD hardware — wanted to actually learn ROCm workflows, model serving, and GPU monitoring in production. Solo submission for the Lablab x AMD AI Academy Challenge.
Submission
Submission draft in progress