Lead AI Engineer — Productivity Systems

Nubank•Palo Alto, CA

4d•Hybrid

About The Position

Nu is seeking an experienced AI engineer to build and deploy LLM-powered systems for internal productivity. The role involves designing agentic workflows, integrating LLMs into business processes, building evaluation and guardrail layers, and transforming manual workflows into AI-assisted ones. This is an applied AI engineering role focused on the AI layer (prompts, context, agents, evaluations, integrations) and product judgment, rather than infrastructure. The engineer will work with frontier models and the modern AI stack, owning the full lifecycle of AI systems from problem discovery to production iteration. Key responsibilities include designing and shipping LLM-powered agents, building evaluation harnesses and guardrails, using orchestration platforms for AI-in-the-loop automation, integrating enterprise platforms, and driving the technical strategy for AI adoption. For Lead/IC6, this includes acting as a technical reference, influencing architecture, mentoring senior engineers, and partnering with security and privacy teams. For Senior/IC5, it involves executing complex AI projects autonomously, identifying automation opportunities, and mentoring mid-level engineers.

Requirements

Shipped LLM systems in production: at least one real system with an LLM at its core — an agent, copilot, RAG application, or AI-driven automation — used by real users, not a proof of concept.
Hands-on AI engineering: practical fluency with prompt and context engineering, tool/function calling, structured outputs, and agent frameworks or orchestration patterns.
Evaluation mindset: experience measuring and improving AI output quality — evals, test sets, feedback loops — and an honest understanding of failure modes (hallucination, drift, prompt injection).
Solid software engineering foundation: proficiency in Python, TypeScript, or Clojure; strong API and integration skills; the discipline to ship maintainable systems, not notebooks.
AI product sense: the judgment to identify which problems deserve an LLM, which need deterministic automation, and which should not be automated at all.
Builder bias: you prototype fast, validate with real users, and harden what works.
Governance-aware: you understand that "efficiency" must be balanced with "security," and you can design AI systems that are safe by default without destroying velocity.
Multiplier: you enjoy documenting your work, creating Golden Paths, and teaching others how to use what you build.
Comfortable with ambiguity: AI capabilities shift monthly; you treat that as an opportunity to re-solve problems better, not as churn.

Nice To Haves

Experience with workflow automation platforms (n8n, Zapier, or custom orchestration engines).
Exposure to cloud services (AWS) and infrastructure-as-code.
Familiarity with AI developer tooling (Claude Code, Cursor, Copilot) and AI governance practices.

Responsibilities

Design, build, and ship LLM-powered agents and workflows that automate complex internal processes end-to-end.
Work hands-on with frontier models and the modern AI stack: tool/function calling, structured outputs, MCP, RAG, multi-agent orchestration.
Own the full lifecycle of an AI system: from problem discovery and prototype to production hardening, monitoring, and iteration.
Build evaluation harnesses, guardrails, and quality feedback loops so AI systems can be trusted in production.
Define what "good" looks like for non-deterministic systems and instrument it: evals, regression suites, human-in-the-loop review where it matters.
Use orchestration platforms (e.g., n8n) and custom integrations as delivery vehicles for AI-in-the-loop automation across business units.
Integrate enterprise platforms (Slack, Google Workspace, Jira, internal APIs) into coherent, AI-assisted workflows.
Drive the technical strategy for AI adoption within engineering and business workflows.
Develop governance frameworks that make AI coding assistants and agents safe, compliant, and effective.
Create Golden Paths, reference implementations, and documentation that let other teams build AI workflows safely on their own.
Act as the technical reference for applied AI in the domain, influence architecture beyond the immediate team, mentor senior engineers, and partner with ITSec and Privacy to align AI solutions with company policy (for Lead/IC6).
Execute complex AI projects with high autonomy, identify workflow bottlenecks worth automating, and mentor mid-level engineers (for Senior/IC5).