S

Siemens

Automation, digitalization, and smart infrastructure provider

AI Native Architect (Enterprise AI Agent Development Platform)

Natural Language Processing & LLMsConfirmed live in the last 24 hours
Full-timeShenzhen, Guangdong Province, ChinaNo salary listedIn-person

About the Job

Requirements

  • A bachelor's degree or higher in Computer Science, Software Engineering, Artificial Intelligence, or a related major.
  • At least 8 years of software development experience, including at least 3 years of AI/LLM application architecture, Agent systems, or related AI engineering practice.
  • Strong capabilities in distributed systems, microservices, cloud-native technologies, and enterprise software architecture, including the ability to independently design complex AI-domain systems and make technical decisions under the chief architect's guidance.
  • Deep understanding of deep learning fundamentals, Transformer architecture and self-attention, ResNet, and batch normalization; deep understanding of LLM application architecture and familiarity with RAG, Tool Calling, Agent Workflow, context engineering, memory management, and model orchestration.
  • Understanding of large-model pretraining, SFT/RLHF, LoRA fine-tuning, model distillation, and quantization; familiarity with AI model testing methodologies and AI application validation systems; ability to build automated model tests, scenario-based benchmark evaluations, and Agent application regression pipelines, and to design model training/fine-tuning pipelines and evaluate model structure and data-distribution impacts on quality, compute, and inference constraints.
  • Practical design or delivery experience with an Agent Runtime or similar intelligent execution system, including long-running tasks, state management, retry and recovery, tool permissions, and control of nondeterminism.
  • Familiarity with at least one mainstream backend language: Python, Go, Java, or TypeScript; ability to validate key architecture designs with code and resolve core technical problems.
  • Familiarity with Docker, Kubernetes, message queues, databases, caches, observability, and common cloud-native technologies; familiarity with GPU resource scheduling and separate training/inference deployments.
  • Experience in AI application evaluation, automated model testing, AI application validation, security governance, performance optimization, and production issue analysis.
  • Ability to design and deliver large-platform architectures is preferred.
  • Experience with model-training/fine-tuning engineering platforms and AI data-governance platform architecture is preferred.

Responsibilities

  • Under the chief platform architect's guidance, plan, design, and evolve the AI Native architecture for the XAgent enterprise AI Agent development platform, establishing an extensible, reusable, secure, and reliable enterprise Agent foundation.
  • Design layered architectures and interface specifications for the AI application layer, Agent Runtime, model and context management, Skill/Tool execution, knowledge retrieval, evaluation and governance, and cloud infrastructure.
  • Define boundaries between AI Native and traditional cloud-platform capabilities; reuse existing identity and access management, multi-tenancy, databases, billing, message queues, and cloud infrastructure to avoid duplicated work.
  • Set AI-domain technology selections, architecture principles, key technical routes, and medium- and long-term evolution plans in response to industrial and enterprise application requirements; submit proposals for the chief architect's platform-wide review.
  • Design Agent task planning, execution loops, state management, long-running tasks, pause and resume, retries, timeout controls, and failure recovery.
  • Design Workflow, single-Agent, and multi-Agent collaboration execution modes, defining their use cases, execution constraints, and engineering boundaries.
  • Establish unified architectures for model integration and routing, inference strategies, context assembly, memory management, and tool calling.
  • Design Skill/Tool registration, discovery, versioning, execution protocols, delegated permissions, runtime isolation, and result validation to support ecosystem partners and third-party tools.
  • Define safe boundaries between AI inference, deterministic business logic, industrial control, and human approval; coordinate cross-module integration proposals with the chief architect.
  • Establish Agent evaluation measures covering task success, answer quality, tool-call accuracy, execution stability, inference cost, and response latency; lead the design of model testing and AI application validation platforms with automated testing for model functionality, accuracy, robustness, adversarial samples, distribution drift, and version comparisons.
  • Build scenario-based test sets, regression cases, and acceptance benchmarks for Agent applications, enabling automated validation and release gates for models and Agent applications.
  • Lead engineering architecture for proprietary and domain models, creating end-to-end pipelines for training-data governance and preprocessing, SFT/LoRA fine-tuning, distillation, quantization, and version management.
  • Review model-structure proposals using foundational principles including Transformer, ResNet, and batch normalization; evaluate model capability boundaries, convergence, memory use, inference latency, and potential defects.
  • Connect training-pipeline outputs to the platform model gateway and Agent Runtime for end-to-end delivery and ongoing iteration of proprietary platform models.
  • Design Agent execution traces, logs, tracing, failure analysis, and production quality monitoring so issues can be located, outcomes traced, and versions regression-tested.
  • Design model and prompt version management, evaluation-set management, continuous integration, and release-quality gates for ongoing AI application iteration.
  • Design enterprise AI security architecture covering prompt-injection protection, data-access controls, tool-call authorization, approval for sensitive operations, sandbox isolation, and audit trails.
  • Establish fallback, degradation, human-intervention, and risk-control mechanisms for model nondeterminism and unstable external dependencies.
  • Design multi-tenant enterprise AI service architecture for unified management of multiple models, Agents, Skills, and business scenarios, with resource isolation and scheduling for training and inference tasks.
  • Optimize model calls, context length, caching, concurrent execution, asynchronous tasks, and resource scheduling to balance success rates, latency, and inference cost; also optimize training data I/O, memory utilization, training duration, and compute cost.
  • Design token-usage collection, execution-task metering, training-compute metering, cost attribution, and observability interfaces, coordinating with existing account, billing, and commercial transaction systems.
  • Integrate the AI Native platform with cloud-native infrastructure to meet high-availability, scalability, operability, and enterprise deployment requirements; submit resource requests and cross-infrastructure integration proposals for chief architect review.
  • Lead technical reviews, AI-domain architecture decisions, and resolution of major technical problems; drive architecture from design into production, submitting cross-domain solutions for the chief architect's final review.
  • Establish AI Native development practices, interface standards, reference implementations, technical documentation, and best practices for model training, model testing, AI application validation, data governance, and model delivery.
  • Guide AI Backend, FDE, and related engineering staff in resolving complex technical issues and establishing consistent implementation and quality standards.
  • Collaborate with product, business, infrastructure, security, and ecosystem teams to translate business needs into actionable technical routes and platform capabilities, coordinating proposals with the chief architect.

Desired Qualifications

  • Experience designing enterprise AI Agent platforms, Agent Runtime systems, AI development platforms, or AI infrastructure platforms.
  • Experience building multi-tenant SaaS platforms, developer platforms, API marketplaces, or third-party ecosystem platforms.
  • Familiarity with MCP, Agent-to-Agent protocols, tool sandboxes, policy engines, or multi-Agent orchestration architectures.
  • Experience with industrial AI, industrial software, OT/IT integration, or enterprise private deployments.
  • Familiarity with model inference services, GPU resource scheduling, inference optimization, model gateways, and multi-model governance.
  • Experience with reliability governance, cost optimization, and security architecture for large-scale production systems.
  • Experience building end-to-end model training, data governance, AI model evaluation, LLM/Agent automated testing, or AI application benchmark validation platforms, including training-data management, dataset versioning, and data-quality governance.

About the Company

Siemens

Automation, digitalization, and smart infrastructure provider

Siemens is a global engineering and technology company focused on automation, digitalization, smart infrastructure, and mobility. Its products and solutions span industrial automation systems, electrical equipment, energy management, and digital services that help industries run more efficiently. The company’s core technology comes from its history of electrical generation and distribution, including the dynamo-electric principle, and it now sells hardware, software, and services that connect factories, buildings, grids, and transportation networks. Siemens differentiates itself through scale, a broad portfolio across the value chain, and a long track record of large-scale, multi-country projects in infrastructure and industry. Its goal is to enable smarter, more sustainable industrial processes and urban systems by integrating hardware, software, and digital services to improve productivity, energy efficiency, and mobility.

Company size
10,001+
Company stage
IPO
Headquarters
Munich, Germany
Founded
1847

Recently Posted Jobs

Explore more opportunities at Siemens

See more jobs