资深数据工程师，机器学习平台
Staff Data Engineer, ML Platform
我们正在寻找一位资深数据工程师，负责构建我们定价和机器学习系统所依赖的数据基础，从原始的外部市场数据到生产环境中做出决策的模型。我们实时从分散的外部市场买入和定价资产。将这些数据转化为模型可以安全消费的形式，并在模型上线后保持其健康状态，是这里最困难的数据问题。你将在精简的数据团队中担任最高级别的数据工程负责人，并负责将数据科学工作从笔记本转移到生产环境。你将负责架构和标准，而不是处理工单队列。
职责描述
- 全程负责外部市场数据的采集、标准化和交付到数据仓库
- 负责核心物品模型和实体解析，将杂乱的输入记录映射到该模型
- 构建质量与监控机制，在利益相关者注意到之前发现数据缺口和性能下降
- 负责数据仓库层（dbt建模、测试、文档），并与产品工程团队建立模式契约，确保上游变更导致构建失败，而非指标失效
- 构建特征和训练数据层：保证时间点正确性、可复现性，并在训练和推理之间保持一致性
- 与数据科学家合作，将模型从原型推进到生产环境——包括批量和实时评分、版本控制、CI/CD以及安全的发布和回滚
- 负责模型可观测性——输入漂移、预测质量、重新训练触发器——确保模型性能下降由警报捕获，而非由错误决策造成
- 将LLM辅助的流程生产化，确保其发挥价值，并内置评估和成本控制机制
- 设定数据团队的工程标准，提升数据科学家的能力水平
查看英文原文
About Courtyard
Courtyard.io is the fastest-growing collectibles startup, ever. We are revolutionizing the world of collectibles trading by enabling instant liquidity and delivering a high-velocity, immersive experience. From trading cards to comics, we're redefining how people discover, collect, and unlock value.
Courtyard is not just another marketplace. All assets are securely vaulted and fully insured, giving collectors peace of mind alongside unmatched speed and simplicity. Whether you're investing, discovering, or curating your dream collection, we've built a platform that's trusted, simple, and built for speed.
And we're just getting started. We're a remote-first company hiring across all functions to push the boundaries of what's possible in collectibles and digital ownership.
Job Summary
Courtyard is hiring a Staff Data Engineer to own the data foundation our pricing and ML run on, from raw external market data all the way to models making decisions in production. We price and buy back assets in real time off a fragmented external market. Turning that into something a model can safely consume, and keeping the model healthy once it's live, is the hardest data problem here. You'll be the senior-most data engineering voice in a lean data org and the person who gets data science work from notebook to production. You'll own architecture and standards rather than a ticket queue.
What You'll Do
- Own external market data ingestion end to end — acquisition, normalization, and delivery into the warehouse
- Own the canonical item model and the entity resolution that maps messy inbound records onto it
- Build quality and monitoring that catches gaps and degradation before a stakeholder notices
- Own the warehouse layer (dbt modeling, tests, documentation) and set schema contracts with product engineering so upstream changes break the build, not the metrics
- Build the feature and training data layer: point-in-time correct, reproducible, and consistent between training and serving
- Partner with data scientists to take models from prototype to production — batch and real-time scoring, versioning, CI/CD, and safe rollout and rollback
- Own model observability — input drift, prediction quality, and retraining triggers — so a degrading model gets caught by an alert, not by a bad decision
- Productionize LLM-assisted pipelines where they earn their keep, with evaluation and cost controls built in
- Set the engineering bar for the data org and level up DS on production-path work
What We're Looking For
- 7+ years across data engineering, ML engineering, or ML platform work with real production ownership, at senior or staff scope
- Experience taming third-party data at volume — unstable sources, schema drift, no advance notice
- Entity resolution in practice, including the judgment to know when a confident match is wrong
- You've shipped ML to production alongside data scientists: built the feature pipelines, deployment path, and monitoring for models that served live decisions, not just handed off a table
- Fluency in where ML breaks in the data: leakage, point-in-time joins, train/serve skew, drift
- Deep SQL and Python; production dbt with tests and incremental models you actually maintained
- Strong dimensional modeling — SCD Type 2, incremental loads, backfills, late-arriving data
- Cloud warehouse depth. We run BigQuery and GCP;
- Comfortable with orchestration, containers, and CI/CD — you can get a service deployed and keep it observable without being an infrastructure specialist
- Clear communicator with strong ownership instincts — you say what you'll own, what you won't, and by when
- Bonus: feature store or model registry experience; LLM evaluation and cost management; pricing or valuation systems; a real interest in trading cards and collectibles
What You'll Get In Return
- A dynamic and engaging environment focused on fostering real growth and innovation
- Opportunities to create amazing products that our customers truly love and value
- Comprehensive health insurance packages with dependent coverage
- Competitive salary with ample opportunities for career advancement and development
- Enjoy the flexibility of a fully remote work environment
- 401(k) plan with a 4% employer match to help you plan for the future
- $100 monthly dogfooding stipend to support trying out our products firsthand
- Access to employee wellness programs designed to support your overall well-being
Originally posted on Himalayas