高级机器学习工程师
Senior Machine Learning Engineer
ZoomInfo 是职业加速的平台。我们行动迅速，敢于大胆思考，并赋予你实现人生最佳工作的能力。你将与 deeply 关心工作、彼此挑战并庆祝胜利的队友们一起工作。借助能放大你影响力的各种工具和一个支持你抱负的文化，你不仅会做出贡献，还会快速实现目标。
你将帮助构建 ZoomInfo AI 代理所依赖的所有智能——关于公司和其中人员的真实信息，它们之间的关系，以及它们正在购买什么。你将从模型学习的数据到其输出所呈现的产品界面，全程负责结果，根据问题需求应用传统的机器学习、统计学或语言模型。你将从数据图谱、实体解析或购买意图中解决一个问题，从设计到生产，然后在发布后负责它。
**你将负责的工作**
- 你将从几乎没有公开足迹的公司网站中提取领导层、地点和产品信息，检测过时记录，根据每份文档的成本选择阅读哪些页面。
- 你将对公开信息稀少的公司进行员工数量和收入估算，使用梯度提升树、缺失输入的回归和校准的不确定性。
- 你将判断两个记录是否描述同一个公司或人，同时衡量错误合并和错误拆分。
- 你将从公司阅读的内容含义中推断购买意图，而不是关键词匹配，处理大量多语言、嘈杂的文本，为倾向评分、类似客户检索和联系人推荐提供支持。
- 你将构建引用来源的研究代理，以及支撑它们的评估系统：人工标注、经过验证的 LLM 评判者和回归门控。
- 你将训练从大型模型中提炼出的小型、任务特定模型，将其与大型模型对比而非完美标准进行评估，并负责其量化和部署。
- 你将从设计到生产解决一个问题，并在发布后负责它，包括其评估和成本，同时通过设计和代码审查帮助其他工程师改进他们的工作。
**你带来的技能**
_必备条件：_
- 你曾将机器学习系统投入生产并负责其发布后的运行，跟踪结果进入所需的任何层级——深度比年限更重要。
- 你具备超越语言模型的传统机器学习专业知识，包括监督学习和大规模数据的特征工程，
查看英文原文
ZoomInfo is where careers accelerate. We move fast, think boldly, and empower you to do the best work of your life. You’ll be surrounded by teammates who care deeply, challenge each other, and celebrate wins. With tools that amplify your impact and a culture that backs your ambition, you won’t just contribute. You’ll make things happen–fast.
You'll help build the intelligence every ZoomInfo AI agent reasons over — what's true about companies and the people in them, how they relate, and what they're buying. You'll own outcomes end to end, from the data a model learns from to the product surface where its output lands, applying classical machine learning, statistics, or language models based on what the problem calls for. You'll take a problem in the data graph, entity resolution, or buying intent from design through production, and own it after launch.
**What You'll Do**
- You will extract leadership, locations, and products from the websites of companies with almost no public footprint, and detect stale records, choosing which pages to read based on cost per document.
- You will estimate headcount and revenue for companies with sparse public information, using gradient-boosted trees, regression with missing inputs, and calibrated uncertainty.
- You will determine whether two records describe the same company or person, measuring both false merges and false splits.
- You will infer buying intent from the meaning of what companies read, not keyword matches, across large volumes of multilingual, noisy text, feeding propensity scoring, lookalike retrieval, and contact recommendations.
- You will build research agents that cite their sources, along with the evaluation systems behind them: human labels, validated LLM judges, and regression gates.
- You will train small, task-specific models distilled from larger ones, measuring them against the larger model rather than against perfection, and own their quantization and serving.
- You will take a problem from design to production and own it after launch, including its evaluation and its cost, while helping other engineers improve their work through design and code review.
**What You Bring**
_Must-Have:_
- You have taken machine learning systems to production and owned them after launch, following the outcome into whichever layer it needs — depth matters more than years.
- You bring classical machine learning expertise beyond language models, including supervised learning and feature engineering on large, messy tabular data, along with applied statistics: experiment design, statistical inference, and calibrated scores under class imbalance.
- You have deployed language processing at scale — text classification, information extraction, and entity linking over large volumes of multilingual, noisy text.
- You have built LLM agents or multi-step systems in production, including tool and context design and failure analysis from traces, and you've evaluated systems with no single right answer using LLM judges validated against human labels.
- You are proficient in production Python and strong in SQL, with experience in distributed data processing, and you use AI coding tools daily, rigorously reviewing their output.
_Preferred:_
- You have experience with ranking and retrieval, including embeddings, learned re-ranking, and metrics such as recall@k, MRR, and nDCG.
- You bring experience in propensity modeling, clustering, or entity resolution on messy, real-world data.
- You have trained and served open-weight models in PyTorch or an equivalent framework, with cost tracked per unit of work.
- You have defended LLM systems against adversarial inputs and prompt injection, or worked on web-scale information extraction, knowledge graphs, or user memory for agents.
#LI-Remote
#LI-VC1
Actual compensation offered will be based on factors such as the candidate’s work location, qualifications, skills, experience and/or training. Your recruiter can share more information about the specific salary range for your desired work location during the hiring process. We want our employees and their families to thrive.
In addition to comprehensive benefits we offer holistic mind, body and lifestyle programs designed for overall well-being. Learn more about ZoomInfo benefits [here](https://www.zoominfo.com/careers#benefits).
Below is the US base salary for this position. Additional compensation such as Bonus, Commission, Equity and other benefits may also apply.
$128,100—$201,300 USD
**About us:**
ZoomInfo (NASDAQ: GTM) is the Go-To-Market Intelligence Platform that empowers businesses to grow faster with AI-ready insights, trusted data, and advanced automation. Its solutions provide more than 35,000 companies worldwide with a complete view of their customers, making every seller their best seller.
ZoomInfo is committed to protecting your privacy when you apply for jobs with us. Please review our Job Applicant [Privacy Notice](https://drive.google.com/file/d/1yiG-k0YX_sW10PiJk_xliDc3W-veJCrK/view?usp=drive_link) for more details on how we handle your personal information.
ZoomInfo may use a software-based assessment as part of the recruitment process. More information about this tool, including the results of the most recent bias audit, is available [here](https://www.zoominfo.com/legal/nyc-local-law-144-notice).
ZoomInfo is proud to be an equal opportunity employer, hiring based on qualifications, merit, and business needs, and does not discriminate based on protected status. We welcome all applicants and are committed to providing equal employment opportunities regardless of sex, race, age, color, national origin, sexual orientation, gender identity, marital status, disability status, religion, protected military or veteran status, medical condition, or any other characteristic protected by applicable law. We also consider qualified candidates with criminal histories in accordance with legal requirements.
For Massachusetts Applicants: It is unlawful in Massachusetts to require or administer a lie detector test as a condition of employment or continued employment. An employer who violates this law shall be subject to criminal penalties and civil liability. ZoomInfo does not administer lie detector tests to applicants in any location.