AI Engineer – Agentic Applications

This is a fully remote position, open to applicants in Brazil.

📋 Description

• Design specialized agents overseen by a supervisor, utilizing agentic loops (ReAct) and maintaining auditable decision logs.

• Create MCP Servers that link agents to operational APIs (.NET) and machine learning models via OpenAPI contracts.

• Develop the MCP Host using LangChain / Semantic Kernel and large language models through Azure OpenAI.

• Establish Human-in-the-Loop (HITL) workflows for critical decision-making processes.

• Guarantee agentic security: unique per-agent identity (OAuth2), tool-level access control, idempotency, prompt-injection sanitation, and prevention of PII leakage.

• Version prompts (PromptOps) and enhance agent observability (LangSmith or similar tools).

• Integrate with event-driven architecture (Azure Service Bus / Kafka).

⛳️ Requirements

• A bachelor's degree in Data Science, Computer Science, Statistics, Engineering, Mathematics, Economics, or a related discipline.

• Proficiency in English or Spanish (for international clients).

• Strong programming capabilities in Python and/or C#/.NET, with a focus on testing, contracts, version control, and code readability.

• Experience in developing LLM applications: including tool/function calling, agent orchestration, prompt engineering, and managing nondeterministic behavior.

• Background in systems integration: designing and consuming REST APIs (OpenAPI/Swagger), OAuth2 authentication, idempotency, and webhooks/callbacks.

• Secure access to LLM-driven data—such as using natural language for queries while implementing scope control, validation, and auditing.

• Knowledge of LLM application security, particularly regarding prompt injection, data leakage, and tool-level permission management.

• Preferred qualifications: MCP (Model Context Protocol), familiarity with LangChain or Semantic Kernel, Azure OpenAI, multi-agent patterns, HITL, LangSmith or similar, Kafka/Azure Service Bus, the .NET ecosystem, and experience in regulated environments (LGPD and auditing of automated decisions).

🏝️ Benefits

• Contractor arrangement (PJ); we are looking for long-term partnerships.

• Opportunity for remote work.

• 8-hour workday (168 hours per month).

• Time bank reconciled every four months, with any surplus hours compensated.

• Variable compensation available after four months of the contract.

• Flexible benefits card: monthly food allowance of R$700.00 on a Flash card.

• Paid time off: 7 consecutive days after one year of the contract; 15 days after two years; 21 days between five and seven years; and 30 days after eight years of the contract.

• Birthday Friday: an additional day off during your birthday month.

• Partnership with Wellhub.

• Partnership with Open English.

• Partnership with FIAP.

• Collaborations with clinics and psychological services.