Staff Software Engineer, Product Workflow Reinvention

About Nu

Nu serves more than 140 million customers, guided by a mission to fight complexity and empower people. The company has been leading an industry transformation through innovative products and human-centered services.

Proprietary technology and data at scale power Nu’s digital platform, built to promote financial access, advancement, and transparency. Its business model thrives on customer love and lower costs, feeding a flywheel of growth and profitability. Visit our Institutional Page

About the role We are hiring a Staff Software Engineer to be the foundational technical lead on the newly established Product Workflow Reinvention team inside Global Products. This is a high-impact, 0-to-1 initiative working directly alongside our CPO structure to shape one of the company’s strategic bets on AI: reinventing how products are designed, specified, and built.

Product Workflow Reinvention aims to replace the current, human-centric approach with an AI-native workflow. As the foundational engineer on this team, you will own the end-to-end technical architecture and build the initial internal platforms that ship weekly improvements to how Nubank builds products.

Mission

Design and operationalize the foundational technical strategy for Product Workflow Reinvention so that:

Intent becomes execution: Product requirements, research, and technical specs flow seamlessly into AI agents and domain execution tools with complete legibility and provenance.

Platform primitives are built for AI-native engineering: Product and platform teams consume well-architected workflows, state machines, and event systems designed specifically for agentic execution.

Every slow step in the life cycle has a tool: Product and engineering teams get first-class tooling for technical discovery, prototype-to-code experimentation, self-serve customer data, service health, and ops automation.

Agentic workflows scale safely, deterministically, and economically: Evaluation harnesses and integration patterns let LLM agents perform complex build tasks reliably, at a defensible cost per outcome, while keeping humans in the loop at strategic decision gates.

The workflow encodes what good looks like, so speed does not come at the cost of rigor.

Core Responsibilities

Lead the architecture of the orchestration layer: Own the end-to-end architecture of the single system of record, including data models, work legibility, and spec delivery to domain agents.

Design AI-native platform primitives: Build new platform abstractions from scratch, applying deep backend expertise to workflows, state management, event orchestration, and agent integrations.

Make agents reliable and affordable: Design agentic workflows, evaluation harnesses, and integration patterns that turn frontier models into practical velocity multipliers. Own the cost side of that equation through context design, model selection, and knowing when a deterministic component beats an inference call.

Ship the first-generation workflow tools: Build and iterate on the internal products that remove the slowest steps in the life cycle, from research and requirements generation to technical discovery, design-to-code experimentation, self-serve customer data, service health, and ops automation. Pick the sequence based on where teams actually lose the most time.

Operate in a high-autonomy, greenfield environment: Translate broad product vision and monthly milestones into concrete architectural decisions, implementation roadmaps, and sequencing.

Core Qualifications

Senior Staff-Level Engineering Experience: Proven track record designing, building, and operating complex backend or platform systems at scale.

End-to-End Architectural Ownership: Demonstrated ability to set technical direction, write comprehensive design docs/RFCs, and establish primitives that other engineers build against. Expertise in data modeling, workflow engines, state machines, and event-driven architecture.

0-to-1 Platform Delivery: Experience leading ambiguous platform initiatives from early discovery to launch and rapid iteration. Comfort operating in small, fast-moving environments by creatively combining APIs, open-source tools, and AI models rather than over-engineering.

Built for other builders: Experience building internal tools or platforms whose users were other product creators, with evidence you drove real adoption rather than only shipping capability. You understand why teams abandon an internal tool in week two. Bonus: you have worked on a customer-facing product that also became a platform other products were built on.

AI and agentic conviction: Strong, well-reasoned conviction around LLM agents, agentic systems, or spec-driven development, with real examples of using AI tools, automation, or decision-support mechanisms to accelerate development workflows. You are connected to where the field is moving without being carried by the hype, and you can name concretely what agents cannot yet do reliably.