Senior AI Harness Engineer

About the role

The Senior AI Harness Engineer joins the client's AI development team to define and build the foundational engineering capabilities supporting AI models and agents at enterprise scale. The role reports to senior technical leadership and establishes engineering standards, solves complex systemic issues and partners with multiple technical teams to deliver enterprise-grade AI solutions. It is an on-site position, initially scheduled for 2 weeks starting 17 September 2026.

Key responsibilities

• Define the architecture and set technical direction for reusable AI harness capabilities.
• Build reusable SDKs, libraries and reference implementations for AI system development.
• Lead the design of interfaces connecting AI agents to enterprise data and operational systems.
• Develop evaluation strategies covering safety, reliability, latency and costs.
• Establish instrumentation and observability standards for AI system diagnostics.
• Resolve reliability and systemic issues through technical investigations.
• Partner with internal teams to establish and uphold engineering standards and practices.
• Evaluate emerging AI infrastructure and frameworks for potential adoption or standardisation.

Required profile

• Right to work in the intended location.
• Degree in Computer Science, Engineering, AI/ML or a related discipline, or equivalent hands-on experience.
• At least 8 years of experience in software engineering, platform development, distributed systems or AI/ML with production system operation.
• Ability to influence engineering decisions as a senior technical authority.

Required skills

• Python, APIs, distributed systems, concurrency, testing and system design.
• Hands-on experience architecting and deploying production LLM, RAG or agentic AI systems.
• Enterprise security protocols including identity, authentication and authorisation, secrets management and privilege enforcement.
• Cloud-native architectures and distributed systems.
• Advantageous: knowledge of agent frameworks such as LangGraph or LangChain.