Distributed SystemsRubyGoAIMLLLMLarge Language ModelsOpenAIAnthropicRAGAgenticLeadership
About this role
Role Overview
Lead AI Strategy & Execution: Drive the roadmap for our conversational AI stack, moving beyond simple decision trees into LLM-driven reasoning, RAG, and agentic workflows.
Orchestrate the AI Ecosystem: Oversee the integration of third-party AI solutions while simultaneously scaling our in-house LLM infrastructure to handle high-stakes crypto support queries.
Build Evaluation & Guardrails: Establish rigorous AI evaluation frameworks (LLM-as-a-judge) and feedback loops to ensure our models are accurate, grounded, and compliant with global financial regulations.
Agentic Automation: Move from "chat" to "action" by building secure pathways for AI agents to perform complex tasks (e.g., transaction troubleshooting, account recovery) via internal APIs.
Drive Technical Architecture: Define how we handle vector databases, prompt engineering, and context window management to provide a personalised experience for every Coinbase user.
Operational Excellence: Own the reliability of AI services, including latency optimisation, cost management (token usage), and fallback mechanisms to human agents.
Requirements
8+ years of software engineering experience, with 2+ years leading high-performing teams in a fast-paced environment.
Hands-on AI/ML Leadership: Proven experience shipping products powered by Large Language Models (LLMs). You understand the nuances of prompt engineering, fine-tuning, and the current landscape of model providers (OpenAI, Anthropic, etc.).
Systems Thinking: Experience building RAG (Retrieval-Augmented Generation) pipelines and managing the data lifecycle required to ground AI in real-time knowledge.
Platform Mindset: You’ve built scalable, distributed systems and understand how to integrate AI components into a high-traffic production environment (Go, Ruby, or similar).
Evaluation Obsessed: You don’t just "vibe check" AI; you have experience with quantitative evaluation frameworks to measure hallucination rates, accuracy, and customer sentiment.
Security & Safety First: A deep understanding of how to build AI "guardrails"—ensuring models don't leak PII or hallucinate financial advice.
Tech Stack
Distributed Systems
Ruby
Go
Benefits
Total compensation may also include equity and bonus eligibility and benefits (including medical, dental, and vision)