About PiecePiece is a holding company that acquires vertical software/AI businesses and scales them through creator-powered distribution. We believe the next Latin American giants will emerge at the intersection of AI and creator economy — software brings technology, recurring revenue, and LTV; creators bring audience, trust, and distribution. Piece brings them together.Our edge is a proprietary creator-focused distribution engine that uses AI to automate growth across our vertical software portfolio — social media, sales funnels, content, and creator operations. Every vertical we add makes the machine smarter, compounding proprietary data on what works in creator-led distribution for AI/Software. Our goal is to hit US$10M ARR in 2026.We're building Piece to be a home for some of the most exceptional people in the world. Talent density is non-negotiable, the bar is unreasonably high, and we expect excellence in everything we ship. Our ambition is to put Brazil on the map as a global protagonist in growth and distribution.The RoleWe're looking for a Senior Backend Engineer to be the technical pillar of the product at the first and main company in our portfolio — someone who combines technical depth with product sense, and who will own the architecture, evolution, and operation of its entire backend and AI infrastructure.You'll be hands-on building and scaling the systems that orchestrate our agents, process messages in real time, and sustain the platform's accelerated growth. This isn't a maintenance role — it's a building role.If you enjoy solving complex problems in distributed systems, have a genuine interest in applied AI, and want to build something that's reshaping how the Brazilian legal market operates, this is the place. You'll join as part of our founding team, with competitive compensation and meaningful equity through stock options.What you'll doBackend architecture and evolutionDesign and evolve the microservices that power the platform, with a focus on scalability and maintainabilityMake technical decisions autonomously — database modeling, API design, stack choicesRaise the team's technical bar through code review and mentorshipAI agents and LLMsBuild and evolve our agent pipelines: orchestration, tool calling, dynamic workflows, and intelligent routingOptimize cost and latency of LLM calls without compromising qualityEvolve our RAG, document processing, and voice synthesis featuresIntegrations and operationsEnsure reliability of asynchronous message flows and external integrationsBuild new integrations as the product evolvesMaintain operational stability in a multi-tenant system with paying customersQuality and observabilityEnsure end-to-end traceability across synchronous and asynchronous flowsStructure operational metrics and maintain visibility over agent behavior in productionIdentify and resolve performance bottlenecks before they become incidentsWhat we expect from youBackend & SystemsProficiency with Node.Js + TypeScript in high-throughput production systems — including API design, modular architecture, and long-term maintainability decisionsStrong grasp of event-driven and asynchronous architectures: message queues, pub/sub patterns, idempotency, retry logic, and failure isolation in distributed systemsExperience designing multi-tenant SaaS backends: data isolation strategies, per-tenant configuration, and operational complexity at scaleSolid database modeling — relational (schema design, indexing, query optimization) and NoSQL (document modeling, consistency trade-offs) — and ability to choose the right store for the problemHands-on cloud experience (GCP preferred): Cloud Run, Pub/Sub, Firestore/Cloud SQL, IAM, and observability toolingAI & Agent SystemsExperience building LLM-powered pipelines in production — not just calling APIs, but thinking through orchestration, context management, fallback behavior, and output reliabilityUnderstanding of agent architectures: tool calling, multi-step reasoning, routing logic, state management across turns, and how to make agents predictable in real-world conditionsPractical knowledge of RAG pipelines: chunking strategies, embedding selection, retrieval quality, and reranking — and awareness of where RAG breaks downCost and latency awareness in LLM usage: prompt engineering for efficiency, model selection trade-offs, caching strategies, and when not to use an LLMFamiliarity with evaluation and observability for AI systems: how to measure agent quality, detect regressions, and maintain visibility over non-deterministic behavior in productionProduct & Technical JudgmentAbility to reason about product trade-offs, not just technical ones — understands what to build, what to cut, and whyComfortable operating with ambiguity: capable of going from a fuzzy problem to a concrete technical proposal without needing a fully-specced ticketHas opinions about prioritization and isn't just an executor — questions requirements upstream and thinks about user and business impact before writing the first line of codeUses AI tools natively in the development workflow (coding assistants, automated testing, code review, documentation) and stays current with the evolving tooling landscape. Understands that development velocity is a competitive advantage, not just an output metricQuality & Operational MaturityAbility to instrument distributed systems end-to-end: structured logging, distributed tracing, alerting, and SLO thinkingExperience operating asynchronous flows with real customers: understanding of what breaks silently and how to catch it before they doComfort with ambiguity — able to define the right technical approach when the problem itself is still being figured outNice to havePython for ML/data pipelines or LLM experimentationGo for latency-critical or high-throughput servicesExperience with LLM fine-tuning, RLHF, or systematic model evaluationBackground in regulated or compliance-heavy domains (legaltech, fintech)What we offerFounding team seat — real ownership of what gets built and howCompetitive compensationMeaningful stock optionsOn-site environment with high talent density and direct exposure to foundersDetailsWork model: On-site (São Paulo or São José dos Campos)Availability: Full-timeContract: PJ