Resume — Alex Chen【免费下载链接】career-opsOpen-source AI job search: scan job portals, evaluate listings into a structured A-H report with a global 1-5 score, tailor your CV, track applications — runs locally in your AI coding CLI (Claude Code, Codex, OpenCode, Antigravity…)项目地址: https://gitcode.com/GitHub_Trending/ca/career-opsLocation:Austin, TXEmail:alexexample.comLinkedIn:linkedin.com/in/alexchenPortfolio:alexchen.devGitHub:github.com/alexchenProfessional SummaryFull-stack AI engineer with 6 years building production ML systems. Led the ML platform at a Series B fintech (2020-2024), scaling from 2 models to 15 in production. Built real-time fraud detection (99.7% precision, $2M/year saved), recommendation engine (18% uplift), and an internal MLOps platform serving 4 engineering teams.Work ExperienceTechFin Corp -- Austin, TXSenior ML Engineer / ML Platform Lead2020-2024Led ML platform team (3 engineers), built internal MLOps tooling: model registry, A/B testing framework, feature storeDesigned real-time fraud detection pipeline: Kafka → feature computation → model inference → decision engine. 99.7% precision at 50ms p99Built recommendation engine for lending products: collaborative filtering LLM reranking. 18% conversion upliftReduced model deployment time from 2 weeks to 4 hours with CI/CD pipeline (GitHub Actions SageMaker)Established model monitoring: drift detection, performance dashboards (Grafana), automated retraining triggersDataStartup Inc -- RemoteML Engineer2018-2020Built NLP pipeline for document classification (BERT fine-tuning, 94% accuracy on legal docs)Implemented search ranking with learning-to-rank modelsSet up experiment tracking with MLflow and model versioningProjectsFraudShield(Open Source) — Real-time fraud detection framework. Kafka Streams feature store model serving. 500 GitHub starsLLM Eval Toolkit(Open Source) — Evaluation framework for LLM applications. Supports custom metrics, regression testing, CI integrationEducationMS Computer Science, UT Austin (2018)BS Computer Science, UC Berkeley (2016)SkillsML/AI:PyTorch, TensorFlow, scikit-learn, Hugging Face, LangChainMLOps:SageMaker, MLflow, Kubeflow, Airflow, Feature StoreInfra:Kubernetes, Kafka, Redis, PostgreSQL, AWSLanguages:Python, Go, TypeScript, SQL## 逐节拆解这个结构为什么长成这样 ### 联系信息块加粗键名 冒号 纯值 markdown **Location:** Austin, TX **Email:** alexexample.com联系信息用加粗键名 冒号的 Markdown 惯例五个字段Location / Email / LinkedIn / Portfolio / GitHub覆盖了简历评估与后续申请流程需要触达候选人的最小集合。注意样例刻意用--双连字符而非 em-dash 书写公司名与城市TechFin Corp -- Austin, TX以及用普通连字符写年份区间2020-2024——这与仓库的 ATS 归一化约定一致下文会专门讲。Professional Summary三段式浓缩全部可验证摘要不是自我评价而是三句带数字的简历预告片总量声明——6 年生产级 ML 系统经验规模声明——从 2 个模型扩到 15 个生产模型结果声明——99.7% 精度、每年节省 $2M、18% 转化提升、服务 4 个工程团队。这一写法对应 examples/ats-normalization-test.md 中被明令禁止的套话清单——passionate about machine learning、results-oriented professional with a proven track record、5 years of experience in artificial intelligence 等空洞表述在生成内容中从一开始就不该出现。对比之下样例摘要里的每个形容词背后都挂着一个可复核的数字。同时AGENTS.md 对写作有硬性的事实约束任何量化声明必须能溯源到主文件cv.md、config/profile.yml、modes/_profile.md等衍生文件如面试准备文档中的数字不能反哺为主事实。因此写摘要时先保证这些数字在你自己经历里真实存在才值得写进来。Work Experience标题块 全量化 bullet每段经历的排版遵循固定四元组###三级标题 公司名 城市/远程标记加粗行 职位**Senior ML Engineer / ML Platform Lead**独立一行 时间段2020-2024其下是 bullet 列表。看 TechFin Corp 的 5 条 bullet可以提炼出 career-ops 示范的证明点句式bullet结构证明点带领 3 人团队构建 MLOps 工具链职责 交付物model registry、A/B testing framework、feature store实时反欺诈管道架构 指标Kafka → 特征计算 → 模型推理 → 决策引擎99.7% 精度 50ms p99贷款产品推荐引擎技术路线 业务指标协同过滤 LLM 重排18% 转化提升部署时间从 2 周降到 4 小时前后对比CI/CDGitHub Actions SageMaker模型监控体系持续性工程漂移检测、Grafana 面板、自动重训触发每一条都是做了什么 用了什么 结果量化的三段式没有一条停留在负责……相关工作这种无法被 ATS 关键词匹配、也无法被评估模型判分的描述。第二条 bullet 里的→箭头是刻意的它用 ASCII 可读的方式表达管道流向而在导出 PDF 时会被归一化为英文介词见下文 ATS 一节。Projects开源项目承担作者身份证明- **FraudShield** (Open Source) — Real-time fraud detection framework. ... 500 GitHub stars - **LLM Eval Toolkit** (Open Source) — Evaluation framework for LLM applications. ...Projects 在 resume 变体中只保留两个开源项目各带一条能力声明 技术栈 可验证规模。这里要特别注意 AGENTS.md 中的红线作者身份声明不可谈判——除非cv.md或article-digest.md明确把某个项目归属于你任何下游流程都不允许声称你 author 了某个仓库/框架/工具用过 X 不等于 造了 X。因此样例把开源项目放在 Projects 而非 Experience 里让归属关系一目了然。Skills按维度分组而不是平铺关键词- **ML/AI:** PyTorch, TensorFlow, scikit-learn, Hugging Face, LangChain - **MLOps:** SageMaker, MLflow, Kubeflow, Airflow, Feature Store - **Infra:** Kubernetes, Kafka, Redis, PostgreSQL, AWS - **Languages:** Python, Go, TypeScript, SQL技能按 ML/AI、MLOps、Infra、Languages 四个维度分组并加粗前缀。这种分组不是美观问题而是下游工具链的输入格式upskill.mjs 会从cv.md/config/profile.yml提取已知技能来做加权技能差距图jd-skill-gap.mjs 是零 LLM 的 JD 技能分类器对照cv.md输出 existing / supportedByResume / gap 三态且永远不会自动往cv.md里加声明。技能条目越具体Kubeflow而非分布式系统这些离线工具的可匹配性越高。ATS 友好书写为什么样例避开了花哨 Unicodeexamples/ats-normalization-test.md 是 generate-pdf.mjs 文本归一化通道的回归夹具列出了所有会导致 ATS 解析错误或乱码的 Unicode 码点名称码点归一化结果Em-dashU2014-En-dashU2013-弯引号双/单U201C/D、U2018/9/省略号U2026...零宽空格等U200B 等删除不换行空格U00A0普通空格而实现位于 generate-pdf.mjs 的sanitizeText函数中它在上面之外还处理了两类 PDF 文本提取器容易丢失的字符// Arrows often stripped by PDF text extractors — replace with ASCII for ATS safety. t t.replace(/\s*\u2192\s*/g, () { bump(right-arrow, 1); return to ; }); t t.replace(/\s*\u2190\s*/g, () { bump(left-arrow, 1); return from ; }); // Middle dot and bullet glyphs garble in some extractors — replace with pipe. t t.replace(/\s*\u00B7\s*/g, () { bump(middot, 1); return | ; }); t t.replace(/\s*\u00B7\s*/g, () { bump(middot, 1); return | ; });见 generate-pdf.mjs。每次替换都会计入统计生成 PDF 时日志会打印类似 ATS normalization: N replacements (em-dashX, smart-double-quoteY, ...)的行——examples/ats-normalization-test.md 末尾给出了端到端验证命令node generate-pdf.mjs /tmp/dirty-cv.html /tmp/clean-cv.pdf --formata4。回到样例简历Kafka → feature computation → model inference → decision engine这条 bullet 在 PDF 导出后会变成Kafka to feature computation to model inference to decision engine在 ATS 纯文本检索中依然能命中 Kafka、inference 等关键词。这说明两件事在源 Markdown 里用→是安全的导出层会兜底归一化但用--写年份区间和公司分隔、避免弯引号和 em-dash 则是从源头零替换的最稳写法归一化只修字符不修文风——examples/ats-normalization-test.md 明确写道 The normalizer does NOT fix writing style空洞套话属于写作期就被 modes/_shared.md 规则禁止的内容。career-ops 如何消费这份文件从模板到流水线输入定位与归属User Layer 的规范事实源在 career-ops 的分层模型里cv.md属于User Layer永不自动更新的个人层与config/profile.yml、modes/_profile.md并列AGENTS.md 将其列为个性化数据的核心。AGENTS.md 进一步明确项目根目录的cv.md是规范 CV。也就是说examples/resume-example.md只是模板你基于它写出的根目录cv.md才是整个系统的事实源。首次使用doctor 检查 结构化建档examples/README.md 给出的落地路径只有四步npm run doctor检查前置条件以cv-example.md或本文件的 resume 变体为结构指南写出cv.md可选用 article-digest-example.md 作为article-digest.md的模板能提升评估质量职业跨越两个典型画像如工程师 讲师时参考 examples/dual-track-engineer-instructor/ 的完整配置。AGENTS.md 定义了cv.md缺失时的建档流程Agent 会请求你粘贴现有简历、LinkedIn 信息或口述经历然后从你提供的任何材料创建cv.md——干净的 Markdown带标准分区Summary、Experience、Projects、Education、Skills。这份样例的五级结构Professional Summary/Work Experience/Projects/Education/Skills正是该流程的目标形态。分区顺序有守卫不能随意调序cv.md的分区标题不是装饰导出 HTML/PDF 时其顺序会被校验。tests/cv-section-order.test.mjs 中固定的期望顺序是const expected [Professional Summary, Work Experience, Projects, Skills, Certifications, Education];【免费下载链接】career-opsOpen-source AI job search: scan job portals, evaluate listings into a structured A-H report with a global 1-5 score, tailor your CV, track applications — runs locally in your AI coding CLI (Claude Code, Codex, OpenCode, Antigravity…)项目地址: https://gitcode.com/GitHub_Trending/ca/career-ops创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考