技术负责人，机器学习
Technical Lead, Machine Learning
职位描述
目前有超过50亿用户在使用基础应用，如电子邮件、笔记、任务、日历等，他们并不具备AI原生能力。我们的使命是为全球任何不习惯复杂提示的用户打造主动式应用。我们旨在通过最少甚至无需提示的方式，将智能带入对话、杂务、组织和工作流程中。
我们的产品专注于实现长期运行工作流的高可靠性、持久上下文和现实任务完成。我们认为产品将大大减少幻觉现象
我们的目标是整理每个人的生活，让我们都能把时间花在有价值和有意义的事情上
作为机器学习技术负责人，您将负责我们智能系统的执行层，将研究和模型能力转化为可靠、可扩展的生产系统。
您将在整个模型生命周期中工作：数据、训练、评估、推理和部署。这是一个需要亲力亲为的领导角色，适合希望在研究、系统和产品交汇点上工作的人员。
您将负责
- 负责支撑公司所有ML系统的端到端架构，从数据和训练到评估、推理和部署。
- 构建和演进大型模型的训练和微调管道。
- 设计衡量能力、稳健性、安全性和实际产品性能的评估系统。
- 架构高性能推理系统，优化延迟、GPU利用率、内存、成本和可靠性。
- 构建用于高质量真实世界和合成训练数据的数据管道和系统。
- 建立可靠的生产基础设施，用于部署、监控和持续改进模型。
- 与研究和应用工程团队紧密合作，将模型能力转化为产品改进。
- 根据实际性能做出务实的技术权衡并快速迭代。
我们寻找的人选
- 具备构建和交付用于生产的ML系统经验，而不仅仅是研究原型。
- 对现代大模型训练、微调、评估和推理有深刻理解。
- 具备扎实的软件工程和系统基础。
- 具备在有意义规模下运行ML工作负载的经验，尤其是基于GPU的系统。
- 具备强大的技术判断力，并能独立解决模糊问题。
- 倾向于实验、测量和交付。
- 对正确性、可靠性和生产质量有高标准。
成果
- 研究
查看英文原文
About the Role
There are over 5 billion users using basic applications today such email, notes, tasks, calendar and they're not AI-native. Our mission is to build proactive applications for anyone in the world, who are not used to complex prompting. We aim to bring intelligence to conversations, errands, organising and workflows, with minimal to no prompting.
Our product focuses on achieving high reliability for long-running workflows, persistent context, and real-world task completion. We believe products will greatly reduce hallucinations
Our objective is to organise anyone's life, allowing us all to spend time on valuable and meaningful things
As Technical Lead, Machine Learning, you own the execution layer of our intelligence, turning research and model capabilities into reliable, scalable production systems.
You will work across the model lifecycle: data, training, evaluation, inference, and deployment. This is a hands-on leadership role for someone who wants to operate at the intersection of research, systems, and product.
What You'll Own
- Own the end-to-end ML systems powering our company, from data and training to evaluation, inference, and deployment.
- Build and evolve training and fine-tuning pipelines for large models.
- Design evaluation systems that measure capability, robustness, safety, and real-world product performance.
- Architect high-performance inference systems, optimizing latency, GPU utilization, memory, cost, and reliability.
- Build data pipelines and systems for high-quality real-world and synthetic training data.
- Establish reliable production infrastructure for deploying, monitoring, and continuously improving models.
- Partner closely with research and application engineering to turn model capabilities into product improvements.
- Make pragmatic technical trade-offs and rapidly iterate based on real-world performance.
What We're Looking For
- Experience building and shipping ML systems used in production, not just research prototypes.
- Strong understanding of modern large-model training, fine-tuning, evaluation, and inference.
- Strong software engineering and systems fundamentals.
- Experience operating ML workloads at meaningful scale, particularly GPU-based systems.
- Strong technical judgment and the ability to navigate ambiguous problems independently.
- A bias toward experimentation, measurement, and shipping.
- High standards for correctness, reliability, and production quality.
Outcomes
- Research and models reliably translate into production-ready solutions with clear performance and quality targets.
- ML pipelines, training loops, and inference systems are stable, efficient, and maintainable.
- Production issues are detected, debugged, and resolved quickly, minimizing user impact.
- Team members are supported, aligned, and able to deliver high-impact ML work with minimal friction.
- Iterations on models and systems are measurable, safe, and improve user experience over time.
Tech Stack
- Python
- PyTorch / JAX
- GPU-based training and inference system
Ideal Experience
- You have built or shipped real ML systems used by people, not just demos.
- You are comfortable working with large models and understanding their failure modes.
- You write strong, production-grade code and care about system correctness.
How We Work
We are a small, high-talent-density, hands-on team. Engineers have broad ownership and are expected to exercise strong judgment and execute independently.
We make decisions quickly, work closely together, and balance speed with engineering fundamentals. We care less about process and more about building something exceptional.
Interview process
If there appears to be a fit, we'll reach to schedule 3, but no more than 4 interviews.
Applications are evaluated by our technical team members. Interviews will be conducted via virtual meetings and/or onsite.
We value transparency and efficiency, so expect a prompt decision. If you've demonstrated the exceptional skills and mindset we're looking for, we'll extend an offer to join us. This isn't just a job offer; it's an invitation to be part of a team that's bringing AI to have practical benefits to billions globally.
Originally posted on Himalayas