Liepin needed to identify where large models fit into recruiting workflows and then turn them into measurable gains in talent matching, interviewing, cost and hiring efficiency.
Reported impact
Liepin turned large-model capabilities into a scalable interview and matching product that reduces hiring effort, improves evaluation consistency and broadens the commercial reach of its talent platform.
Liepin and Baidu AI Cloud focused their large-model work on talent selection, where structured evaluation and high manual effort create strong automation value. The resulting capability is Doris, an AI-driven interview product that combines Liepin's recruiting expertise with Baidu's large-model stack. Instead of only matching resumes to jobs, Doris evaluates candidates across more than one hundred dimensions from the iceberg model, runs eight kinds of tests in one process, supports customized interview creation in four steps and provides multilingual candidate interaction with a human-like experience. The system is designed to reduce the amount of manual work required from interviewers while making evaluation more standardized and scalable. In effect, Baidu's models become the intelligence layer for candidate screening, interview execution and report generation, while Liepin supplies the recruiting domain model, test logic and HR-service distribution network. That makes the product useful across enterprise hiring, probation evaluation, talent inventory and internal mobility workflows.
Candidate and job data enter Liepin's Doris workflow, where Baidu-backed models run tests, interviews and scoring logic, then generate structured outputs for hiring, probation and talent-management decisions.
At scale, enterprises can save more than 90% of interview costs and improve execution cycles by 60%-70% (original: 规模应用后可以，为企业节约面试成本90%以上……平均面试执行周期较传统面试提升60%—70%).
Baidu's large-model capabilities now deeply empower Liepin's Doris product (original: 目前百度智能云的大模型能力已深入赋能猎聘·Doris产品).
Cheating detection reaches 95% accuracy (original: 全方位智能防作弊，作弊检测准确率达95%).
Doris covers more than 100 evaluation dimensions and achieves over 90% scoring consistency with senior interviewers (original: Doris覆盖冰山模型七个层级的100多个评估维度，面试评价与资深面试官评分一致性大于90%).
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Baidu AI Cloud Sample Dec 29, 2025 Zhaopin Zhaopin needed to move beyond isolated AI features and make large models useful across the full recruiting workflow, from job posting through candidate screening, communication, interviewing, and matching decisions. HR operating steps were reduced 80%. DeepSeek-R1 · ERNIE Speed · Qianfan AppBuilder Baidu AI Cloud Sample Jun 19, 2025 Achievo (Beijing Teamsun Technology Co., Ltd.) Achievo wanted to upgrade its traditional intelligent-customer-service stack, which struggled with intent errors, weak context handling and repetitive answers, even though the company already had deep industry knowledge… Intelligent Data Assistant · Intelligent Speech Engine · Multi-Turn Dialogue Engine 
OpenAI Sample Oct 15, 2026 Sullivan & Cromwell Sullivan & Cromwell needed to review draft agreements against relevant precedents, client priorities, and market practice in one workflow so lawyers could do deeper legal analysis without losing transparency into… Agreement Analyzer · ChatGPT Enterprise 