Jaime Cernuda
Back to Publications
eScience'262026Conference

Acropolis: A Portable, AI-Driven Semantic Search System for HPC

Rajni Pawar, Luke Logan, Jaime Cernuda, Xian-He Sun, Anthony Kougkas
Semantic SearchAI AgentsMCPHPC

Cite this work

@inproceedings{pawar2026acropolis,
  title={Acropolis: A Portable, AI-Driven Semantic Search System for HPC},
  author={Pawar, Rajni and Logan, Luke and Cernuda, Jaime and Sun, Xian-He and Kougkas, Anthony},
  booktitle={22nd IEEE International Conference on eScience (eScience 2026)},
  year={2026}
}

Acropolis augments a tiered filesystem with adaptive-depth content indexing and exposes it to LLM agents through a typed natural-language interface. A transparent file-summary operator runs inside the I/O ingest path, a pluggable backend abstraction surfaces lexical, vector, and graph search engines, and an MCP server exposes the search surface to any compatible agent.