Succeeding at Scale: Automated Multi-Retriever Fusion and Query-Side Adaptation for Multi-Tenant Search
Prateek Jain, Shabari S Nair, Ritesh Goru, Prakhar Agarwal, Ajay Yadav, Yoga Sri Varshan Varadharajan, Constantine Caramanis · 2026
Read Succeeding at Scale: Automated Multi-Retriever Fusion and Query-Side Adaptation for Multi-Tenant Search by Prateek Jain, Shabari S Nair, Ritesh Goru, Prakhar Agarwal, Ajay Yadav, Yoga Sri Varshan Varadharajan, Constantine Caramanis (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-04646.txt.
Description
Prateek Jain and six co-authors make company search adaptable without rebuilding it, in a 2026 arXiv paper. Large multi-tenant search systems sit on mountains of unlabelled query logs, while retraining both the query and document encoders would force re-indexing thousands of isolated tenant corpora. Operationally out of the question. The unused logs are dark data: valuable, and locked.
Their contributions come in two parts. First, DevRev-Search, a passage retrieval benchmark for technical customer support. Built by a fully automated pipeline that fuses sparse and dense retrievers for candidates and uses a language model judge for consistency filtering and relevance labels. Second, a systematic study of index-preserving adaptation: finetuning only the query encoder, documents and indices untouched. Parameter-efficient tuning of that encoder alone, they find, delivers a striking quality-per-efficiency trade-off on their benchmark and two public ones.
Catalog record
| CREATOR | Prateek Jain, Shabari S Nair, Ritesh Goru, Prakhar Agarwal, Ajay Yadav, Yoga Sri Varshan Varadharajan, Constantine Caramanis |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-04646