Agent Harness Engineer

About Manus AI

Manus AI offers a versatile AI agent designed to bridge the gap between human intention and task execution, efficiently managing intricate workflows that include deep research, data analytics, and software development. This technology empowers users to delegate entire complex workflows seamlessly.

Role Overview

We seek a skilled engineer to develop foundational systems for the agent runtime and control layers. Your responsibilities will span managing context, tools, execution, feedback, and safety boundaries, with accountability for designing and delivering robust components that enhance reliability in real-world tasks.

Key Responsibilities

• Take ownership of essential components such as context and memory systems, model and tool orchestration, task planning and execution, integration with tools and MCP, and mechanisms for evaluation and feedback.
• Identify problems and define success metrics independently when faced with ambiguous requirements or technical challenges; lead technical decisions, design, implementation, deployment, and ongoing improvements.
• Enhance agent capabilities by refining context engineering, prompt engineering, retrieval processes, tool usage, and system architecture, focusing on measures such as task success rate, reliability, latency, and cost efficiency.
• Develop consistent agent evaluation methodologies, regression testing frameworks, and end-to-end observability tools using actual task data and production failure analysis to drive enhancements.
• Collaborate closely with product, research, infrastructure, and security teams on design reviews, risk assessments, delivery, and enhancement of engineering quality.

Candidate Requirements

• Demonstrate a solid foundation in software engineering and computer systems with a minimum of five years’ professional experience or equivalent proficiency, proven by delivery of maintainable and testable production-grade software.
• Experience in backend or distributed systems including concurrency, task queues, data persistence, idempotency, failure recovery strategies, and observability practices.
• Capability to independently manage complex systems or critical components; exercise sound technical judgment and maintain control over complexity to ensure delivery of quality outcomes.
• Effective use of AI tools to support research, coding, and verification while maintaining autonomous judgment, alongside proactive communication and collaborative skills.
• Competence in collaborating on technical work using Mandarin and literacy in English for routine technical communication.
• Preferred experience includes production involvement with agent runtimes, harness systems, multi-step workflows, task schedulers, browser automation, continuous integration runners, or code execution platforms.
• Additional desirable expertise includes deep knowledge of context engineering, long-term memory management, retrieval and ranking methods, model routing, tool utilization, MCP, multi-agent frameworks, or agent evaluation processes.
• Familiarity with execution isolation, adhering to least privilege principles, multi-tenancy architecture, defenses against prompt injections, high-risk action approval frameworks, auditing practices, open-source contributions, or technical writing is favored.

What Manus AI Offers

• The opportunity to work at the forefront of AI agent technology within a fast-paced, innovative team, transforming cutting-edge AI advances into practical applications.
• An equity-sharing plan giving employees a stake in Manus AI’s long-term success and value creation.
• Unlimited access to Manus tokens that empower you to experiment, build, and maximize productivity.

Additional Information

If you are driven to work with emerging technologies and desire to impact the AI landscape, we encourage your application.