Senior AI Engineer

Fluxo is a startup focused on digitizing the used-car market. We are building, from scratch, a platform that makes buying and selling used cars safe end to end — including custody of the deal's funds until vehicle delivery. The company is AI-native from the start: AI drives the closing — it reads documents, maps fraud risk, resolves exceptions, and keeps the parties moving — while money only moves under auditable rules.

We are looking for a Senior AI Engineer to work side by side with the Founder and the Tech Lead, building the AI layer that allows the operation to scale without growing at the same rate as the team — always within clear security boundaries, since the product handles third-party funds and sensitive data. This position comes with equity.

What you will do

• Build document reading and data extraction for deal documents (CRLV, CNH, inspection report, contract), with cross-checking among them.
• Detect fraud risk in documents and deals: altered documents, inconsistent data, known scam patterns.
• Build agents that conduct conversations with buyers, sellers, and intermediaries (including WhatsApp), where each party sees only what they are entitled to see.
• Automate the handling of operational exceptions and turn each exception resolved by a person into a system rule or test case.
• Prioritize, together with the Founder, what to automate first, based on measured time for each task since the first deal.
• Create guardrails with clear limits of action: AI never moves money or alters financial records on its own — sensitive actions go through system rules and human approval.
• Integrate automations into the product engine (TypeScript, PostgreSQL), working with the Tech Lead, who is responsible for the overall architecture.
• Set up continuous evaluation (evals) and production observability: quality, latency, human intervention rate, and AI cost per deal, with a defined cap.
• Protect automations against prompt injection and personal data leakage (LGPD).

What we are looking for

• A solid background in software engineering, with experience putting AI/LLM systems into production for real users — not just prototypes.
• Experience with document data extraction in production (multimodal LLMs, OCR, structured output).
• Experience with agent orchestration (LangGraph, LangChain, CrewAI, Semantic Kernel, or equivalents) — the framework matters less than judgment about when to use an agent, a classic model, or a simple rule.
• Practical experience with RAG and vector databases (pgvector, Pinecone, Weaviate, or similar).
• Implementation of guardrails and AI security: output validation, limits on automated action, human approval, and prompt injection prevention.
• Experience with evaluation (evals) and observability of AI systems in production, including cost control per transaction.
• Experience automating operational business processes (not just chatbots or content assistants).
• Ability to work directly with the Founder, translating business needs into technical solutions.
• Experience in a startup in the building phase (greenfield).
• Advanced English.

Nice to have (not disqualifying):

• Fraud detection or risk models in production
• Classic machine learning (classification, anomaly detection, ranking)
• Recommendation or matching systems
• Experience with products that handle financial transactions or sensitive data
• LLMs in Portuguese and Brazilian documents
• Customer service and communication automation via WhatsApp Business API
• TypeScript/Node.js (our main stack) in addition to Python
• PostgreSQL and Supabase
• Fluent use of AI tools in development (code assistants, agents)

Work model: Hybrid in São Paulo

Contractor (PJ) arrangement: Compensation + Equity