高级/资深机器学习工程师（模型开发）
Senior/Staff Machine Learning Engineer (Model Dev)
我们致力于通过人工智能技术变革癌症治疗方式。我们开发了基础模型，用于分析临床和病理数据，生成可操作的见解，指导治疗方案选择并改善癌症患者的预后。通过持续改进这些模型，我们希望揭示驱动癌症进展的生物学机制。
我们正在寻找一位经验丰富的机器学习工程师，负责AI生物标志物的全流程开发——从与临床和生物统计学合作伙伴共同定义问题，到模型开发和验证，再到监管提交和生产部署。除了负责生物标志物项目外，您还将解决我们领域中最困难的跨领域问题：在不同扫描仪和机构间的鲁棒性、模型决策的机制可解释性，以及我们病理学基础模型的下一代版本。
平等就业机会：在Artera，我们重视汇聚来自不同背景的个体，为患者和医生开发创新解决方案。作为一家平等就业机会的雇主，我们不会基于种族、肤色、宗教、国籍、年龄、性别（包括怀孕）、身体或精神残疾、健康状况、基因信息、性别认同或表达、性取向、婚姻状况、退伍军人保护状态或其他受法律保护的特征进行歧视。
核心职责：
- 与产品、生物统计学、临床开发和法规/质量团队合作，主导技术工作并定义面向患者的产品的战略愿景。
- 设计并构建基于AI的生物标志物，利用多模态数据——包括全切片图像、临床变量和分子数据——预测患者预后、治疗获益和分子特征。
- 推进我们的核心自监督基础模型以及建立在其上的下游架构（多实例学习、生存时间/风险模型、分割和分类组件），以泛化能力作为首要目标。
- 确保在不同扫描仪、机构、染色协议和患者群体中的评分可重复性。
- 开发并集成机制可解释性方法，以解释模型决策，建立临床医生信任，并推动可操作的模型改进。
- 构建工具和流程，简化端到端的模型开发生命周期——从原型设计到生产部署和监控。
查看英文原文
About Us: Artera is an artificial intelligence company dedicated to transforming cancer care. We’ve developed foundation models that analyze clinical and pathology data, generating actionable insights that guide therapy selection and improve outcomes for cancer patients. By continuously improving these models, we aim to uncover the biological mechanisms driving cancer progression.
We're looking for an experienced machine learning engineer to own AI biomarker development end to end — from problem framing with clinical and biostatistics partners, through model development and validation, to regulatory submission and production deployment. Beyond owning a biomarker program, you'll take on the hardest cross-cutting problems in our field: robustness across scanners and sites, mechanistic interpretability of model decisions, and the next generation of our pathology foundation models.
Equal Employee Opportunity:At Artera, we value bringing together individuals from diverse backgrounds to develop new andinnovative solutions for patients and physicians. As an equal opportunity employer, we do notdiscriminate on the basis of race, color, religion, national origin, age, sex (including pregnancy),physical or mental disability, medical condition, genetic information gender identity orexpression, sexual orientation, marital status, protected veteran status, or any other legallyprotected characteristic.
Essential Responsibilities:
- Lead the technical effort and define the strategic vision for patient-facing products, in partnership with product, biostatistics, clinical development, and regulatory/quality.
- Design and build AI-based biomarkers on multimodal data — including whole-slide images, clinical variables, and molecular data — to predict patient outcomes, treatment benefit, and molecular traits.
- Advance our core self-supervised foundation models and the downstream architectures built on them (multiple-instance learning, time-to-event / hazard models, segmentation and classification components), with generalization as a first-order objective.
- Own score reproducibility across scanners, institutions, staining protocols, and patient populations.
- Develop and integrate mechanistic interpretability methods to explain model decisions, build clinician trust, and drive actionable model improvements.
- Architect tools and processes that streamline the end-to-end model development lifecycle — from prototyping through production deployment and monitoring — ensuring efficiency, reproducibility, regulatory compliance, and scale.
- Author and defend regulatory and quality documentation, and represent AI in design and development reviews.
- Plan and manage delivery: break multi-quarter programs into milestones, manage dependencies across AI, platform, biostatistics, and clinical teams, surface risk early, and hold submission and launch dates.
- Publish in peer-reviewed journals and present at clinical and ML venues; support external academic and industry collaborations.
- Mentor and coach machine-learning scientists and engineers, fostering their technical growth and collaboration skills, and raise the bar on scientific rigor, code quality, and written communication across the team.
Experience Requirements:
- 5+ years of industry experience building deep learning systems in PyTorch (or TensorFlow).
- 2+ years of experience as a technical lead, launching and monitoring machine-learning products in production environments.
- Demonstrated depth in oncology and biomarker development: familiarity with cancer biology and treatment pathways, clinical endpoints, risk stratification, and what makes a biomarker clinically actionable.
- Demonstrated project management ability — scoping, sequencing, and managing dependencies and risk across multiple teams on dated deliverables.
- Proven ability to communicate complex ML concepts effectively to cross-functional, non-ML collaborators.
- Experience mentoring or managing ML scientists and engineers.
Desired:
- Experience building ML on complex clinical data — medical imaging, multi-omics, or longitudinal patient records — including weakly supervised learning and handling variation across sites, devices, and protocols.
- Experience developing ML in a regulated environment — FDA 510(k)/De Novo, CE/UKCA, SaMD, design controls, or CLIA/LDT validation.
- Experience with self-supervised representation learning (e.g., DINOv3) and adapting medical foundation models to downstream clinical tasks.
- Experience with data from randomized controlled trials and multi-institutional clinical cohorts.
- Peer-reviewed publications and conference presentations; history of external academic or industry collaborations.
- Experience with cloud-scale training and workflow orchestration (e.g., Flyte / Union, Kubernetes, AWS), experiment tracking, and reproducible ML pipelines.