Saved in:
Bibliographic Details
Main Authors: Behdin, Kayhan, Song, Qingquan, Vasudevan, Sriram, Sheng, Jian, Ma, Xiaojing, Zhou, Z, Zhu, Chuanrui, Li, Guoyao, Nguyen, Chanh, Ghosh, Sayan, Sang, Hejian, Baarzi, Ata Fatahi, Ramachandran, Sundara Raman, Wang, Xiaoqing, Lan, Qing, S, Vinay Y, Guo, Qi, Johnson, Caleb, Wang, Zhipeng, Borisyuk, Fedor
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.22101
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915576574640128
author Behdin, Kayhan
Song, Qingquan
Vasudevan, Sriram
Sheng, Jian
Ma, Xiaojing
Zhou, Z
Zhu, Chuanrui
Li, Guoyao
Nguyen, Chanh
Ghosh, Sayan
Sang, Hejian
Baarzi, Ata Fatahi
Ramachandran, Sundara Raman
Wang, Xiaoqing
Lan, Qing
S, Vinay Y
Guo, Qi
Johnson, Caleb
Wang, Zhipeng
Borisyuk, Fedor
author_facet Behdin, Kayhan
Song, Qingquan
Vasudevan, Sriram
Sheng, Jian
Ma, Xiaojing
Zhou, Z
Zhu, Chuanrui
Li, Guoyao
Nguyen, Chanh
Ghosh, Sayan
Sang, Hejian
Baarzi, Ata Fatahi
Ramachandran, Sundara Raman
Wang, Xiaoqing
Lan, Qing
S, Vinay Y
Guo, Qi
Johnson, Caleb
Wang, Zhipeng
Borisyuk, Fedor
contents Large Language Models (LLMs) have demonstrated impressive quality when applied to predictive tasks such as relevance ranking and semantic search. However, deployment of such LLMs remains prohibitively expensive for industry applications with strict latency and throughput requirements. In this work, we present lessons and efficiency insights from developing a purely text-based decoder-only Small Language Model (SLM) for a semantic search application at LinkedIn. Particularly, we discuss model compression techniques such as pruning that allow us to reduce the model size by up to $40\%$ while maintaining the accuracy. Additionally, we present context compression techniques that allow us to reduce the input context length by up to $10$x with minimal loss of accuracy. Finally, we present practical lessons from optimizing the serving infrastructure for deploying such a system on GPUs at scale, serving millions of requests per second. Taken together, this allows us to increase our system's throughput by $10$x in a real-world deployment, while meeting our quality bar.
format Preprint
id arxiv_https___arxiv_org_abs_2510_22101
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Scaling Up Efficient Small Language Models Serving and Deployment for Semantic Job Search
Behdin, Kayhan
Song, Qingquan
Vasudevan, Sriram
Sheng, Jian
Ma, Xiaojing
Zhou, Z
Zhu, Chuanrui
Li, Guoyao
Nguyen, Chanh
Ghosh, Sayan
Sang, Hejian
Baarzi, Ata Fatahi
Ramachandran, Sundara Raman
Wang, Xiaoqing
Lan, Qing
S, Vinay Y
Guo, Qi
Johnson, Caleb
Wang, Zhipeng
Borisyuk, Fedor
Information Retrieval
Machine Learning
Large Language Models (LLMs) have demonstrated impressive quality when applied to predictive tasks such as relevance ranking and semantic search. However, deployment of such LLMs remains prohibitively expensive for industry applications with strict latency and throughput requirements. In this work, we present lessons and efficiency insights from developing a purely text-based decoder-only Small Language Model (SLM) for a semantic search application at LinkedIn. Particularly, we discuss model compression techniques such as pruning that allow us to reduce the model size by up to $40\%$ while maintaining the accuracy. Additionally, we present context compression techniques that allow us to reduce the input context length by up to $10$x with minimal loss of accuracy. Finally, we present practical lessons from optimizing the serving infrastructure for deploying such a system on GPUs at scale, serving millions of requests per second. Taken together, this allows us to increase our system's throughput by $10$x in a real-world deployment, while meeting our quality bar.
title Scaling Up Efficient Small Language Models Serving and Deployment for Semantic Job Search
topic Information Retrieval
Machine Learning
url https://arxiv.org/abs/2510.22101