talent-lens (慧眼)
候选人简历质量评估系统 —— 从高校背景、大厂经历、职业稳定性、资深程度、学历五个维度对候选人进行量化评分并输出综合加权总分。
功能特性
- 五维度评估:高校背景(C9/985/211)、大厂经历(科技/金融/外资科技)、职业稳定性(5年3跳/一年一跳检测)、资深程度、学历
- 多种匹配算法:精确匹配、别名匹配、模糊匹配(子串包含)
- LLM 辅助识别:通过 llmdog 库调用大语言模型,从非结构化简历文本中提取结构化信息
- 灵活配置:YAML 配置文件支持深度合并,可自定义权重、评分规则、高校/大厂名单
- 双接口:CLI 命令行 + Python API
- Rich 终端输出:表格化评估报告,也支持纯 JSON 输出
安装
pip install talent-lens
从源码安装:
git clone https://github.com/talent-lens/talent-lens.git
cd talent-lens
pip install -e .
快速开始
CLI 使用
基本评估
# 使用内联参数
talent-lens evaluate --name "张三" --university 清华大学 --degree 硕士 --company 阿里巴巴 --work-duration 36 --work-duration 24
# 从 JSON 文件加载
talent-lens evaluate --input candidate.json
# 输出 JSON 格式
talent-lens evaluate --input candidate.json --json
# 输出到文件
talent-lens evaluate --input candidate.json --output result.json
候选人 JSON 文件格式
{
"name": "张三",
"universities": ["清华大学", "北京大学"],
"degree": "硕士",
"companies": [
{"name": "阿里巴巴", "duration_months": 36},
{"name": "腾讯", "duration_months": 24}
],
"work_durations": [36, 24],
"total_work_months": 60
}
启用 LLM 辅助识别
# 设置 API Key
export TALENT_LENS_LLM_API_KEY="your-api-key"
# 启用 LLM
talent-lens evaluate --input candidate.json --resume resume.txt --llm
初始化配置文件
# 生成默认配置文件
talent-lens init
# 强制覆盖已有配置
talent-lens init --force
Python API 使用
from talent_lens import Evaluator, load_config
# 加载配置(自动合并默认配置和用户配置)
config = load_config()
# 创建评估器
evaluator = Evaluator(config)
# 候选人信息
candidate = {
"name": "张三",
"universities": ["清华大学"],
"degree": "硕士",
"companies": [
{"name": "阿里巴巴", "duration_months": 36},
{"name": "腾讯", "duration_months": 24}
],
"work_durations": [36, 24],
"total_work_months": 60,
}
# 执行评估
result = evaluator.evaluate(candidate)
# 获取结果
print(f"总分: {result['total_score']}")
print(f"各维度分数: {result['dimension_scores']}")
print(f"高校背景: {result['dimensions']['university']}")
自定义配置
from talent_lens import Evaluator
# 直接传入配置字典
config = {
"weights": {
"university": 30,
"company": 30,
"stability": 15,
"seniority": 15,
"education": 10,
},
"university_scoring": {"c9": 100, "other_985": 85, "other_211": 70, "other": 50, "unknown": 30},
"company_scoring": {"tech": 100, "foreign_tech": 95, "finance": 80, "other": 50, "unknown": 30},
"stability_scoring": {
"levels": {
"very_stable": {"min_months": 36, "score": 100},
"stable": {"min_months": 24, "score": 80},
"moderate": {"min_months": 18, "score": 60},
"frequent": {"min_months": 12, "score": 40},
"very_frequent": {"min_months": 0, "score": 20},
},
"penalties": {"five_years_three_jumps": -15, "one_year_one_jump": -25},
},
"seniority_scoring": {
"levels": {
"senior": {"min_years": 8, "score": 100},
"mid_senior": {"min_years": 5, "score": 75},
"mid": {"min_years": 3, "score": 50},
"junior": {"min_years": 1, "score": 30},
"fresh": {"min_years": 0, "score": 15},
},
},
"education_scoring": {"phd": 100, "master": 80, "bachelor": 60, "associate": 40, "unknown": 30},
}
evaluator = Evaluator(config)
result = evaluator.evaluate(candidate)
配置说明
配置文件
执行 talent-lens init 会在当前目录生成 .talent-lens.yaml,只需修改需要覆盖的配置项:
# 调整权重(总和必须为 100)
weights:
university: 30
company: 30
stability: 15
seniority: 15
education: 10
# 调整高校评分
university_scoring:
c9: 100
other_985: 85
# 启用 LLM
llm:
enabled: true
默认权重
| 维度 | 权重 |
|---|---|
| 高校背景 | 25% |
| 大厂经历 | 25% |
| 职业稳定性 | 20% |
| 资深程度 | 15% |
| 学历 | 15% |
评分规则
| 维度 | 分类 | 分数 |
|---|---|---|
| 高校 | C9 | 100 |
| 高校 | 其他985 | 85 |
| 高校 | 其他211 | 70 |
| 高校 | 未知 | 30 |
| 大厂 | 科技 | 100 |
| 大厂 | 外资科技 | 95 |
| 大厂 | 金融 | 80 |
| 大厂 | 未知 | 30 |
| 稳定性 | 非常稳定(>=36月) | 100 |
| 稳定性 | 稳定(>=24月) | 80 |
| 稳定性 | 一般(>=18月) | 60 |
| 稳定性 | 频繁(>=12月) | 40 |
| 稳定性 | 非常频繁(<12月) | 20 |
| 资深 | 资深(>=8年) | 100 |
| 资深 | 中高级(>=5年) | 75 |
| 资深 | 中级(>=3年) | 50 |
| 资深 | 初级(>=1年) | 30 |
| 资深 | 应届(<1年) | 15 |
| 学历 | 博士 | 100 |
| 学历 | 硕士 | 80 |
| 学历 | 本科 | 60 |
| 学历 | 大专 | 40 |
| 学历 | 未知 | 30 |
稳定性惩罚
- 5年3跳:60个月窗口内跳槽超过3次,额外扣15分
- 一年一跳:连续2段及以上工作经历均不超过12个月,额外扣25分
- 两项惩罚可叠加,扣分后分数不低于0
依赖项
- Python >= 3.9
- click >= 8.0(CLI 框架)
- rich >= 12.0(终端格式化)
- pyyaml >= 6.0(YAML 配置解析)
- llmdog >= 0.1.0(LLM 服务调用)
许可证
MIT License
Metadata
Release files for talent-lens 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| talent_lens-0.1.0.tar.gz | 27.2 kB | Without provenance | Details for talent_lens-0.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| talent_lens-0.1.0-py3-none-any.whl | Python 3 | none | any | Details for talent_lens-0.1.0-py3-none-any.whl |
Total release size: 55.0 kB
Release files / talent_lens-0.1.0.tar.gz
| Download URL | talent_lens-0.1.0.tar.gz |
|---|---|
| Size | 27.2 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
25825ce98a8894484473906bbfb5935551dde1307120fb2ba1a336f5c89eb471
|
|
BLAKE2b-256 checksum How to use checksums |
52de40378cb38c87a04aa8fc1baedd0417f2d6d2bcfd5d1205483e2369afdf12
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.11.3
|
Release files / talent_lens-0.1.0-py3-none-any.whl
| Download URL | talent_lens-0.1.0-py3-none-any.whl |
|---|---|
| Size | 27.8 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
7bab06a8741f3a972e104b8e8f69380fcae938d0f830bdd179d9e89fa962dca8
|
|
BLAKE2b-256 checksum How to use checksums |
99fc27bbb00ccd815680bc062c11e0b272e98def2d5f7f7aebc63ddda252e590
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.11.3
|
Security for talent-lens 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.