Lead Engineer, AI Agent Systems

Lead Engineer, AI Agent Systems

ShanghaiEast – Data & Technology /Full Time /On-siteapply for this jobResponsibilities

Architecture Leadership and Evolution

- Lead the architecture and evolution of next-generation agent infrastructure designed for complex, knowledge-intensive work.

- Define clear boundaries and collaboration mechanisms across three core layers: the execution engine, context and reasoning orchestration, and the agent capability foundation. Ensure high availability, reliability, and long-term extensibility in environments with a low tolerance for hallucinations and incorrect outputs.

Agent Execution Engine

- Design and implement the Agent Loop runtime and its middleware pipelines.

- Lead the execution and orchestration of planning and sub-agent workflows, including task decomposition, dependency management, concurrency control, and execution scheduling.

- Build mechanisms for checkpointing, interruption and resumption, failure recovery, self-healing, authorization, and cost control to ensure the reliable execution of long-running and complex multi-step tasks.

Context and Reasoning Orchestration

- Own the design and implementation of core context orchestration capabilities.

- Develop strategies for input standardization, dynamic capability representation, and hierarchical context-budget management, including structured degradation when resource or context limits are reached.

- Build structured task workspaces that support efficient organization of dynamic context. Address challenges including long-history compression, tool-output normalization, evidence traceability, and the management of information across different stages of a task.

Agent Capability Foundation

Lead the development of foundational agent capabilities, including:

- Secure sandboxed environments using technologies such as Docker, Kubernetes, and AST-based controls

- Multi-layer memory stores

- Retrieval and knowledge-access capabilities

- An MCP (Model Context Protocol) Hub

- Skill execution and management engines

- File-processing and transfer pipelines

- Multi-tenant isolation and security controls

- End-to-end observability and diagnostics

Technical Leadership and Team Enablement

- Remain hands-on and personally contribute code to critical platform modules.

- Lead technical decomposition, architecture decisions, code reviews, and the development of automated evaluation systems and feedback loops.

- Guide the engineering team in translating specific business use cases into reusable platform and infrastructure capabilities.

Qualifications

Engineering and Leadership Experience

- At least five years of professional software engineering experience.

- Proven experience leading the design and delivery of complex software systems beyond standard CRUD applications or basic integrations with AI APIs.

- Demonstrated experience operating as a Tech Lead, Staff Engineer, or equivalent technical leader.

- Experience leading an engineering team of at least three people.

Core Engineering Capabilities

- Strong Python software-engineering skills and the ability to independently own critical platform modules.

- Deep experience with common engineering challenges such as streaming responses, asynchronous and concurrent execution, and multi-model routing and provider integration.

- Strong judgement in balancing system reliability, security, cost, latency, and delivery speed.

- Solid understanding of distributed systems, production architecture, debugging, and operational reliability.

Depth in AI and Agent Systems

Candidates must have substantial hands-on engineering experience with Agent and LLM systems, with deep expertise in at least two of the following three areas:

Execution Engine

- Multi-step reasoning loops

- Tool lifecycle management

- Planning and sub-agent orchestration

- Interruption and resumption

- Failure recovery and self-healing

Context and Reasoning Orchestration

- Input standardization

- Context assembly

- Context and token-budget governance

- Provider-specific request shaping

- Task-stage modelling

- Long-context compression and evidence traceability

Agent Capability Foundation

- Sandbox isolation

- Memory and retrieval systems

- MCP infrastructure

- File-system and file-processing capabilities

- Multi-tenant isolation

- Security, monitoring, and observability

apply for this job