O que a vaga pede
Kinter builds agents for mid-market finance teams. Prepaid, accruals, payroll, reconciliation, and more: the repetitive, judgment-heavy work that keeps finance teams stuck in spreadsheets every close. We're a lean, product-centric team (:30 people) moving fast with an exciting product we launched a few months ago. Here you'll do the best work of your life and have an outsized impact: what you ship today is used by finance teams tomorrow. How You'll Make An Impact Here Evolve the AI harness that powers Kinter's agents: orchestration, tool-calling, context and memory management, and the scaffolding that sits between the models and production accounting workflows Design and ship new agentic capabilities, from new agentic workflows to the structured interfaces that let agents act reliably in a financial system of record Strengthen our human-in-the-loop patterns, confidence thresholds, and fallback behaviors that keep agents correct rather than merely fluent. In this domain, an agent that says "I'm not sure, here's what I'd check" beats one that confidently guesses Expand the evaluation and observability tooling that lets us trust agent behavior at scale: curated datasets, evals, automated regression tests, tracing and logging so we can see why an agent did what it did Work directly with customers and their finance teams to understand where agents can safely act autonomously versus where a human needs to stay in the loop, and translate that judgment directly into the harness Help define Kinter’s agentic architecture as we expand from expense management workflows to a much broader set of agents; the patterns you build now become the template for what comes next Work closely with product, engineering, leadership, and customers directly. Given the size of the team, decisions move fast and your work is visible What You'll Bring To The Team 2+ years of hands-on experience building and shipping autonomous agentic systems in production – not just prototypes, demos, or personal projects 8+ years of experience building and shipping SaaS products. Experience operating agents in a sensitive domain (finance, healthcare, legal, or similar), with a real instinct for why correctness is paramount there. You understand why human-in-the-loop beats a hallucinated best-guess GenAI response, the way it might for something lower-stakes like a marketing email Direct, hands-on experience with agent frameworks (e.g. LangGraph, AgentCore, AutoGen, CrewAI, LlamaIndex) tied to a shipped p