Hi everyone,
I’m an independent researcher seeking an arXiv endorsement for cs.AI (cross-list: cs.SE, cs.CY). This is my first arXiv submission and I have no institutional affiliation, so I need a personal endorsement from an existing arXiv author in a related category.
The paper
“Enhancing Nuclear Safety Through AI-Driven Predictive Quality Methodologies in Document Lifecycle Management”
Presented at the IAEA International Conference on Topical Issues in Nuclear Installation Safety (TIC2026), 29 June to 3 July 2026, IAEA Headquarters, Vienna (Conference CN-345; Session BA.3.1, 1 July 2026).
It introduces a digital “Defence-in-Depth” framework for nuclear technical documentation, splitting AI support into three independent, human-governed layers: Retrieval-Augmented Generation (RAG) for drafting, Context-Augmented Generation (CAG) for independent verification, and an administrative compliance gate. The drafting and verification models are deliberately isolated to prevent confirmation bias, and Human-in-the-Loop authority is preserved at every stage. The AI never has authority to alter the design basis.
In a pilot on public European regulatory data (Loviisa NPP, VVER-440/213), CAG-based auditing (Llama 3.3 70B) flagged high-severity omissions (a missing explicit safety-policy statement and an ASME Boiler & Pressure Vessel Code compliance gap) in under 20 seconds, while RAG-assisted drafting returned traceable, source-cited content in under 1.5 seconds.
Code / orchestration logic: GitHub - Hamad-Alajeel/Enhancing-Nuclear-Safety · GitHub Happy to share the full paper or extended abstract privately before you decide.
To endorse
If you’re an active arXiv author in a related cs category and willing to help, I’d really appreciate it. It takes under a minute, and I’m glad to send the paper first. Thanks for your time,
Jahangir Nasimi