Brazil
SONDA
Keyrus Brazil
AI Engineer – Architect of Intelligence
Brazil
São Paulo, São Paulo, Brazil
Python
Automatic translation. Check the original.
WHO YOU ARE
Why Keyrus, Why Now!
Keyrus is an international group with 2.800 consultants and specialists in 28 countries, built on a conviction: AI does not transform companies. Architected intelligence transforms.
For over 30 years, we have been building the data foundations that make intelligent systems possible, designing the Intelligent Enterprise Operating System, where intelligence is embedded in the core of business processes to create sustainable value: we operationalize intelligence.
AI does not replace humans. It repositions us in a place no system can reach: to understand, decide, create, and innovate.
At Keyrus, you will not just develop skills; you will develop discernment. Your expertise strengthens with every system you design, every business challenge you help solve, and every implementation that evolves from prior learnings.
Over time, you will become one of the rarest professionals of the intelligence era: someone capable of connecting data, AI, and human decision-making at scale, across different sectors, markets, and geographies.
This is not just a position. It is a discipline you master and a story you help write to become a Keyrus Architect of Intelligence.
Technology amplifies capabilities. Keyrus culture differentiates. Excellence in execution connects both.
You are a hands-on, systems-minded engineer who understands that a model is only one part of an intelligent system.
You care about what happens after the model's response: whether the context was correct, whether the agent chose the appropriate action, whether the outcome can be evaluated, and whether the capability remains reliable during its continuous operation in production.
You make technical choices considering trade-offs, without seeking a perfect technology. You use evidence and evaluations to improve systems, communicate limitations clearly, and recognize when a problem belongs to AI Engineering versus another specialty.
You combine curiosity for new capabilities with the engineering discipline required to put them into production.
ABOUT KEYRUS
At Keyrus, we help organizations evolve from experimental AI to industrialized AI, from isolated agents to orchestrated systems, and from insights to execution.
This is the discipline we call Architect of Intelligence: designing the intelligent enterprise operating system, where intelligence is embedded in business processes to create sustainable value.
Driven by our proprietary Human Orchestrated Model™ (HOM), we structure intelligence foundations, human-centric governance, and performance steering to create intelligent environments where technology amplifies human capabilities.
With over 30 years of experience and 2.800 employees across 28 countries, we help organizations go beyond transformation, building adaptable, resilient, and continuously evolving companies.
AI does not transform businesses. Architected intelligence transforms.
Come join the Keyrus Team!
WHAT YOU WILL ARCHITECT
As an AI Engineer, you will design, develop, evaluate, and deploy AI capabilities that enable intelligent systems to retrieve context, reason, interact with enterprise systems, and operate reliably in production.
Your focus will be the intelligence layer of the solution: how models are integrated, how information is retrieved, how agents behave, how quality is evaluated, and how capability functions post-deployment.
You will collaborate closely with Software Engineers, Data Engineers, AI Platform Architects, Field Solution or Deployment Engineers, and Governance specialists, maintaining clear accountability for AI behavior and quality within the solution.
Your responsibilities
- Design and develop AI capabilities for production, including RAG systems, intelligent assistants, and agentic workflows, aligned with the approved solution architecture.
- Design agent behavior concerning instructions, context, tools, state, routing, memory, failure recovery, stopping conditions, and appropriate human validation points.
- Integrate and configure models via APIs or compatible platforms, considering capability, quality, latency, cost, and operational requirements.
- Design evaluations and quality criteria using representative datasets, automated tests, regression tests, task success metrics, answer grounding, and other relevant indicators.
- Deploy, monitor, and enhance AI capabilities in production, using evaluation results, operational evidence, and observed system behavior to guide continuous improvement.
- Implement solution-level reliability and control mechanisms, including validation, guardrails, failure handling, and human oversight, per the project's defined architecture and governance standards.
- Integrate AI capabilities into the overall solution, collaborating with Software Engineers on application interfaces and Data Engineers on information retrieval, data quality, and AI-ready data preparation requirements.
- Document and reuse effective practices, including AI behavior, dependencies, evaluation criteria, limitations, and reusable engineering patterns that strengthen future Keyrus projects.
- Enable client teams to operate and expand agents post-production.
WHAT YOU BRING
Experience and Qualifications
- Solid hands-on experience engineering generative AI / LLMs in production, beyond work confined to notebooks or demos.
- Strong Python and software engineering fundamentals, including APIs, testing, version control, code review, and sustainable production code.
- Hands-on experience designing and improving RAG and information retrieval systems, including context preparation, retrieval quality, grounding, and relevant data dependencies.
- Hands-on experience engineering AI agents or multi-step AI workflows, including tool use, context management, routing, state, failure handling, and appropriate human control.
- Hands-on experience creating evaluations for LLMs or agent systems, using measurable criteria and representative test cases to assess behavior and detect regressions.
- Experience deploying and operating AI capabilities in production, with awareness of monitoring, reliability, latency, costs, and continuous improvement.
Desirable Requirements
- Experience integrating commercial or open-source model providers and understanding trade-offs between models and deployment options.
- Experience integrating AI capabilities with enterprise applications, APIs, data sources, or workflow systems.
- Hands-on knowledge of guardrails, validation, human oversight, and responsible AI engineering practices at the solution level.
- Experience with at least one major cloud or AI ecosystem, with the ability to apply engineering principles across comparable platforms.
- Familiarity with containerization, CI/CD, observability, and MLOps/LLMOps practices to collaborate effectively with Software and Platform Engineers.
- Ability to optimize AI systems considering quality, latency, context usage, model selection, and inference cost.
- Connection to business outcomes: relating agent behavior to the business KPI it serves.
- Integration with client systems: CRM, ERP, MCP, and webhooks.
Differentiators
- Deep experience in agentic AI, including multi-agent orchestration, planning and execution loops, complex state management, retrieval strategies, and advanced human-in-the-loop patterns.
- Experience with multimodal systems, fine-tuning, model adaptation, open-source model deployment, or specialized inference.
- Experience with classic machine learning or deep learning, where relevant to the project.
- Experience in regulated environments or those with high security requirements.
- Experience in consulting or professional services.
- Experience mentoring other engineers or creating reusable patterns, components, or evaluation approaches for AI Engineering.
- Experience with low-code platforms for agents, such as n8n, Make, and Dust, for rapid automations complementary to code development.
WHAT WILL MAKE YOU SUCCEED
- You use evaluations and production evidence to guide engineering decisions.
- You balance quality, reliability, latency, cost, safety, and maintainability.
- You systematically diagnose unexpected AI behaviors, rather than treating model output as a black box.
- You understand the limits of your technical autonomy and involve the appropriate specialty when decisions belong to architecture, software, data, governance, or delivery.
- You communicate technical choices and trade-offs clearly and pragmatically.
- You turn successful engineering patterns into reusable knowledge that improves future deliveries.
OUR VALUES
Collective Intelligence
Collaboration across different specialties, roles, and geographies allows us to combine knowledge and deliver more complete answers to our clients' challenges.
Reliability
The ability to deliver complex projects with rigor, quality, and consistency is a cornerstone of our relationship with clients and partners.
Pragmatism
We prioritize concrete impact and measurable value over abstract technological discourse.
Entrepreneurial Spirit
Curiosity, energy, autonomy, and initiative are the foundation of our culture and drive continuous innovation.
For more information: keyrus.com.br
We wish you good luck! 😊🚀
More opportunities with one profile
Want to be considered for more roles without repeating the whole process? Create your profile and be found by companies using Nort.
Create my profile