高级机器学习运维工程师
Senior Machine Learning Operations Engineer
### 关于我们
Hungryroot 正在利用人工智能打造有史以来最以消费者为中心的食品和健康公司。我们就像您的健康生活助手——了解您的目标、生活方式和预算，为您和您的家人推荐并配送健康的杂货、简单的食谱和必要的补充剂。
这是最简单的方式来吃得健康、实现目标、节省时间并发现新食物。我们认为食物是健康的基石，便利不应该意味着妥协，而且每个人在饮食和生活方式上都是独特的。这就是我们正在构建的未来：让健康生活既容易又愉快。
Hungryroot 是一个分布在 28 个以上美国州的顶尖人才团队。虽然我们总部位于纽约市，但我们的远程优先文化强调协作、团队建设和灵活性。您将期待定期的虚拟团队活动、强大的所有权和责任感，以及年度公司聚会。
### 关于该职位
我们正在招聘一名高级机器学习运维工程师加入 Hungryroot 的数据科学团队。我们的团队负责支撑 Hungryroot 客户杂货推荐和箱体个性化的生产系统。
我们的平台结合了 Python 服务、运行在 AWS 上的 FastAPI API、Databricks 上的 Spark 流水线，以及为实时决策引擎提供支持的机器学习模型。该系统正在积极演进，我们正在投资工程基础架构，以使其能够随着业务扩展和适应。
您将与数据科学家、运筹研究人员和产品工程师紧密合作，构建可靠且可扩展的模型驱动个性化系统。这是一个塑造 Hungryroot 客户体验核心部分架构的机会。
### 职责
- 设计、构建和运营可扩展的后端服务、API 和数据流水线。
- 提升生产 ML 和优化系统的可靠性、性能和可观测性。
- 负责从训练好的模型到生产的路径：模型版本控制和注册（MLflow）、安全的发布和回滚，以及数据质量和模型漂移的监控。
- 构建干净的接口，使新的 ML 模型和决策能力能够安全高效地集成，包括实验和功能标志工具。
- 强化不断增长的代码库中的工程基础：自动化测试、类型检查、CI/CD、基础设施即代码、文档和周到的系统设计。
- 推动工程实践的持续改进，确保系统可维护性和可扩展性。
查看英文原文
### **About Us**
Hungryroot is using AI to build the most consumer-centric food and wellness company to ever exist. We act as your personal assistant for healthy living—getting to know your goals, lifestyle, and budget, and recommending and delivering healthy groceries, easy recipes, and essential supplements for you and your family.
It’s the easiest way to eat healthy, achieve your goals, save time, and discover new foods. We believe food is the foundation of health, convenience should not mean compromise, and that everyone is unique in how they eat and live. That’s why we’re building a future in which healthy living is both easy and enjoyable.
Hungryroot is a distributed team of top talent across 28+ U.S. states. While we have a headquarters in New York City, our remote-first culture emphasizes collaboration, team-building, and flexibility. Expect regular virtual team events, strong ownership and accountability, and an annual company retreat.
### **About the Role**
We’re hiring a Senior Machine Learning Operations Engineer to join Hungryroot’s Data Science team. Our team owns the production systems that power grocery recommendations and box personalization for Hungryroot customers.
Our platform combines Python services, FastAPI APIs running on AWS, Spark pipelines on Databricks, and machine learning models that feed a real-time decisioning engine. The system is actively evolving, and we’re investing in the engineering foundations that will let it scale and adapt with the business.
You’ll partner closely with data scientists, operations researchers, and product engineers to build reliable, extensible systems for model-driven personalization. This is an opportunity to shape the architecture behind a core part of Hungryroot’s customer experience.
### **Responsibilities**
- Design, build, and operate scalable backend services, APIs, and data pipelines.
- Improve the reliability, performance, and observability of production ML and optimization systems.
- Own the path from trained model to production: model versioning and registry (MLflow), safe rollout and rollback, and monitoring for data quality and model drift.
- Build clean interfaces that let new ML models and decisioning capabilities integrate safely and efficiently, including experimentation and feature-flag tooling.
- Strengthen engineering foundations across a growing codebase: automated testing, type checking, CI/CD, infrastructure as code, documentation, and thoughtful system design.
- Profile data-heavy services and pipelines; reduce execution time and memory footprint where it matters.
- Collaborate with data scientists, operations researchers, and product engineers to translate business needs into robust technical solutions.
### **Qualifications**
- 5+ years in MLOps, ML engineering, or DevOps with a focus on production ML infrastructure.
- Strong Python and SQL; Bash for automation and tooling.
- Experience designing and operating backend services and APIs (e.g., FastAPI) with attention to reliability, latency, and scalability.
- Hands-on experience with Databricks and Spark (jobs/workflows, Unity Catalog a plus) and MLflow or comparable model lifecycle tooling (registry, versioning, experiment tracking).
- Experience building CI/CD for ML or data systems (Git, GitHub Actions/Jenkins, Databricks Asset Bundles) and infrastructure as code (Terraform or similar).
- Solid AWS fundamentals: IAM, networking, compute/cluster management, containerized workloads (Docker; ECS or EKS).
- Experience with production observability: metrics, logging, alerting, and ML-specific monitoring like data quality and model drift
**Nice to Haves**
- Familiarity with recommendation, personalization, or operations research systems — especially productionizing them.
- Experience with optimization solvers and OR tooling (e.g., Gurobi, OR-Tools) alongside data science or operations research teams.
- Experience integrating experimentation and feature-flag platforms (e.g., Statsig) into production ML services and data pipelines, ideally with warehouse-native setups on Databricks.
- Feature store experience (Databricks Feature Store, Feast, Tecton) serving consistent online/offline features.
- Experience with low-latency model serving and deployment patterns (canary, blue/green, shadow).
- Experience optimizing cost and performance of data-heavy workloads (Spark tuning, cluster right-sizing).
- Additional languages such as Scala or C++.
### **Perks & Benefits**
- Remote-first: work from home, work from our NYC office, work from anywhere in the U.S. - you decide!
- Equity
- Unlimited vacation policy
- Universal paid parental leave
- Monthly Hungryroot credit for delicious, healthy groceries
- Comprehensive health, vision, dental, and life insurance
- 401k with Company Match
- A work from home stipend to support your initial home-office setup
_Expected Pay Range_
_$170,000 - $210,000_
#LI-REMOTE
**_The employer will not sponsor applicants for work visas._**
_Our mission to help make healthy eating easy, accessible, and joyful is better served by a diverse workplace. We are a proud Equal Opportunity Employer committed to building an inclusive workplace. We have zero-tolerance for harassment or discrimination. We do not discriminate on the basis of any protected class._