Research / reproducible public work
Questions tested through artifacts and evidence.
Peer-reviewed conference proceedings and arXiv preprints on high-energy computational physics and low-latency LLM caching architectures.
Research timeline
Evidence accumulated over time.
Publication activity grouped by recorded year, without treating preprints as peer-reviewed work.
- 0120262recorded
Published record
Research stays connected to what can be reproduced.
Each entry preserves its authors, maturity, date, and available public surfaces. A preprint is not presented as peer review.
2026 / arXiv preprint · arXiv:2609.05463 [cs.DC], PollinationsAI Research
A Three-Layer Caching Architecture for Low-Latency LLM Web Search on Commodity CPU Hardware
Ayushman Bhattacharya* and Nihal Gazi* (*Equal Contribution)