Ma Liu Shui / Engineering - Software

Job Title: Multi-Agent Collaboration Engineer

Job Responsibilities

• Join the echOS Agent Core team, responsible for the engineering implementation of multi-agent systems in real-world transaction scenarios, focusing on industrial-grade applications rather than simulation or theoretical verification environments.

• Design and implement role division and responsibility boundaries for multi-business agents, covering core business modules including pricing, inventory, fulfillment, channels, planning, and more.

• Build an agent lifecycle management mechanism to support dynamic behaviors such as startup, suspension, resumption, and reorganization, ensuring system flexibility and stability.

• Define agent permission boundaries and duty segregation mechanisms to ensure module autonomy and system security.

• Develop inter-agent communication protocols and state broadcasting mechanisms to achieve cross-agent state synchronization and data consistency.

• Build a conflict detection and negotiation framework to resolve business goal conflicts such as price vs. inventory, timeliness vs. cost, and support learnable negotiation decisions.

• Integrate centralized and distributed hybrid decision-making architectures, and adopt Multi-Agent Reinforcement Learning (MARL) to continuously optimize overall system revenue.

• Build a multi-agent simulation and replay environment to support strategy validation and anomaly reproduction.

• Develop a collaborative behavior observability system, including decision chain tracing, responsibility attribution analysis, and quantitative effect evaluation.

• Build an A/B testing framework to scientifically evaluate the actual impact of different collaboration strategies.

• Design open access protocols to support the rapid integration of new business modules (e.g., promotion agents) into the existing collaboration network.

Job Requirements

• Bachelor’s degree in Software Engineering, Computer Science, Artificial Intelligence, Automation, Mathematics, or related fields.

• Proficient in Python with solid programming and engineering practice experience; experience with PyTorch is a strong plus.

• Experience in developing complex systems, distributed systems, or agent systems is preferred.

• Understand basic concepts of reinforcement learning or game theory, and able to translate theoretical tools into practical collaboration mechanisms.

• Capable of decomposing complex business logic into multiple autonomous modules, and conducting system-level abstraction and architecture design.

• Keen insight into system degradation, goal conflicts, strategy drift and other issues in real industrial scenarios, and willing to solve them in depth.

• Able to communicate daily in Mandarin and Cantonese; good written communication and cross-team collaboration skills.

• No strict experience requirement; interns and fresh graduates are welcome. Priority will be given to candidates with any of the following backgrounds:

— Experience in Multi-Agent Reinforcement Learning (MARL) projects

— Research or practice related to game theory or mechanism design

— Experience in scheduling systems or resource allocation systems

— Experience in enterprise distributed systems or high-availability service architecture

Benefits

• Competitive remuneration

• Year-end bonus

• Marriage leave