AI Engineer – Trust & Explainability (AI Platform)

About this role

Build the tracing, evaluation, and explainability layer for MeridianLink's shared AI agent runtime. You'll create instrumentation and tooling that lets engineers understand multi-agent workflows end-to-end and enables product teams to surface trustworthy explanations to customers in the lending space.

What you'll do

• Implement multi-agent tracing across gateway, orchestration, memory, and tool layers to track agent actions and decisions
• Design and build correlation logic linking actions across multiple agents including handoffs, branches, and retries
• Integrate and extend open-source observability and explainability frameworks for LLM systems
• Develop evaluation harness components with golden datasets, rubric-based scoring, and CI-integrated quality checks
• Build customer-facing explanation primitives appropriate for regulated lending contexts
• Contribute to platform guardrails and tenant isolation tests ensuring security and data protection

What they're looking for

• LLM-based systems and agent architectures
• Tracing, observability, and instrumentation design
• Python or similar backend language
• Open-source LLM frameworks (LangChain, LlamaIndex, or similar)
• Testing and evaluation methodologies for non-deterministic AI output
• Security and threat modeling for AI applications
• System design and component architecture
• Code review and technical communication