Staff GenAI Engineer

Responsibilities:

• GenAI Strategy and Integration: Define best practices and scalable patterns for integrating LLMs across products and engineering teams.
• Accuracy and Safety: Implement verification mechanisms, human-in-the-loop workflows, and guardrails to ensure AI outputs do not become unverified sources for financial decisions.
• Evaluation and Testing: Build evaluation frameworks, golden datasets, regression tests, and CI quality gates to measure performance and prevent regressions.
• Model and Vendor Strategy: Select appropriate models, optimize token costs, and design flexible architectures that avoid vendor lock-in.
• AI Solution Design: Evaluate when to use LLMs versus deterministic, rule-based approaches to ensure reliability, efficiency, and accuracy.
• Production Reliability: Monitor AI system performance, detect model drift and quality degradation, and establish effective incident detection and response mechanisms.

Requirements:

• 8+ years of engineering experience, with 2+ years building production Gen AI systems that real users depended on
• You have shipped a system at a meaningful scale and been on the hook when it broke. It does not have to be a Gen AI system; distributed systems judgment transfers, and we would rather have it than not
• Real agentic experience: tool calling, multi-step reasoning, agents acting against live business systems. Not a prototype
• Strong RAG fundamentals: retrieval quality, chunking strategy, embeddings, evaluation, and knowledge system design. "We used a vector database" is not sufficient
• Evidence of setting a technical standard that teams outside your own adopted, without having authority over them
• Comfort operating as a senior IC with org-wide scope rather than a team to manage

Nice to have:

• Financial systems, billing platforms, revenue operations, accounting, or fintech background.
• MCP, LangGraph, LiteLLM, or equivalent gateway work, and Langfuse/OpenTelemetry for LLM observability.
• Experience with sandboxed code execution, context compaction, or multi-tenant AI architecture.
• Early-stage or 01 product experience.