Highlights

Set the applied science roadmap for the pod across the core AI capabilities the product depends on, for example multimodal perception, language and agentic reasoning, time series modeling, control. Design and own the evaluation and benchmarking frameworks for generative and predictive models, including offline metrics, online experimentation, and robustness testing in industrial conditions.

Description

Senior Principal Applied Scientist

Job ID

502622

Posted since

09-Sep-2026

Organization

Data & Artificial Intelligence

Field of work

Research & Development

Company

Siemens Ltd., China

Experience level

Experienced Professional

Job type

Full-time

Work mode

Office/Site only

Employment type

Fixed Term

Location(s)

Beijing - Beijing Shi - China
Shanghai - Shanghai Shi - China
Shenzhen - Guangdong Sheng - China

The role

Siemens builds the systems the physical world runs on: factories, power grids, buildings, trains, hospitals. Industrial and physical AI is a major opportunity in applied AI, and one of the harder ones to get right. There is a generation of AI-powered products to build.
We are forming an applied science organization in China to build the science behind them. The work sits at the intersection of machine learning research, real world data, and production systems running in industrial environments.
As Senior Principal Applied Scientist, you set the scientific direction for the pod. You decide which problems are worth solving with ML, what methods to apply, how to evaluate them, and how to move them from a research result to a model that runs in production. You are accountable for the science: the rigor, the evidence, and the outcomes.
This is a senior individual contributor role. Your impact comes from owning the hypotheses, the evaluation, and the path from research to production, and from raising the scientific bar across the team.

Key responsibilities

Set the applied science roadmap for the pod across the core AI capabilities the product depends on, for example multimodal perception, language and agentic reasoning, time series modeling, control.
Convert product and system requirements into clear research questions, hypotheses, and success metrics.
Design and own the evaluation and benchmarking frameworks for generative and predictive models, including offline metrics, online experimentation, and robustness testing in industrial conditions.
Lead applied research projects end to end, from literature review and method selection through experimentation, ablation, and productization
Work with engineers to take models into production grade pipelines: data readiness, optimization, inference, observability.
Influence architectural and system decisions with scientific evidence and tradeoff analysis.
Identify and de-risk scaling challenges: data quality, model drift, latency, throughput, cost, safety.
Mentor scientists and engineers on experimentation rigor, reproducibility, and documentation.
Champion responsible and trustworthy AI: bias detection, model risk management, human in the loop controls.

Basic qualifications

10+ years in applied machine learning, AI research, or data science, with a track record of models that shipped to production and made an impact.
Strong foundation in machine learning theory and practice across training, evaluation, and deployment.
Demonstrated experience setting the science direction for a portfolio of work and shipping it through to production with engineering teams.
Proficiency in Python and modern ML frameworks and toolchains.
Track record of building evaluation and benchmarking that the team can run a roadmap against.
Clear written and verbal communication, with the ability to explain complex ML concepts to engineers, product managers, and senior leaders

Preferred qualifications

Experience setting science direction across multiple capability areas at the same time.
Experience bringing applied research into real world products in industrial or physical domains: manufacturing, automation, robotics, energy, mobility, infrastructure, healthcare.
Breadth across multimodal ML, generative AI, retrieval, agentic workflows, control, or planning.
Experience standing up evaluation, online experimentation, or production monitoring as practices, not just for a single model.
Publications, patents, open source contributions, or significant internal technology transfers that the field can point to.
Experience hiring and mentoring senior or staff level scientists.
Experience working with globally distributed research, product, or engineering organizations.