Senior Agentic Ai Engineer (Pcs786)

Overseas; Rio de Janeiro; Remote; BRL 599,000 - 898,000; Full time

Benefits offered by this job

Paid in USD

Paid Time Off

Holidays observed

Remote work

US business hours

Ownership of AI architecture

Enterprise customers

Job summary

Overseas is seeking an experienced software engineer to own and architect production-grade agentic AI systems. You will design multi-agent orchestration, context handling, and model routing, working with enterprise customers to deliver scalable AI solutions.

The role emphasizes ownership, deep technical leadership, and collaboration across teams, with a fully remote setup and US business hours alignment.

Qualifications

• 5+ years of professional software engineering experience.
• Hands-on production AI systems experience.
• Strong written and verbal English communication.

Responsibilities

• Design production-grade agentic and multi-agent systems.
• Architect agents with defined roles, tools, context, permissions, and decision boundaries.
• Design orchestration and routing across agents, tools, models, and workflows.
• Build systems using multiple LLMs and model providers based on task needs.
• Design context, memory, and knowledge systems for agents.

Skills

Agentic AI design

Multi-agent systems

LLM integration

System orchestration

Model routing

English communication

Tools

Claude Code

Codex

Cursor

Agent SDKs

Neo4j

PostgreSQL

Job description

Why You'll Want to Join

• You will be paid inUSD(bi-monthly: every 15th and 30th)
• Paid Time Offin accordance with company policy
• Observance ofHolidaysper company guidelines
• 100% remote setupso you can work wherever you're most productive
• This role requires availability duringUS business hours
• High ownership over AI architecture and technical direction at an early, high-leverage stage
• Work directly with enterprise customers and see the real impact of what you build

What You'll Work OnAgentic AI Architecture

• Design and build production-grade agentic and multi-agent systems
• Architect specialized agents with clearly defined roles, tools, context, permissions, and decision boundaries
• Design orchestration and routing strategies across agents, tools, models, and workflows
• Build systems that intelligently use multiple LLMs and model providers depending on the task
• Determine which models should power specific agents based on quality, reasoning capability, reliability, latency, and cost
• Design failure handling, fallbacks, guardrails, and human-in-the-loop mechanisms

Agent and Context Graph Interactions

• Design and manage how agents interact with context graphs, knowledge graphs, memory systems, and other context sources
• Determine what information and which portions of a context graph individual agents should be able to access
• Design context retrieval, filtering, ranking, and permission strategies based on each agent's role and task
• Build and evaluate interactions between agents and nested or interconnected context graphs
• Determine how much context an agent needs without unnecessarily increasing tokens, latency, or noise
• Design systems that dynamically provide agents with the most relevant context for a specific task
• Evaluate how changes to context availability affect agent accuracy, reasoning, reliability, and performance

AI Evaluation and Reliability

• Design sophisticated evaluation frameworks for agentic AI systems
• Build evals across model, agent, and context graph interactions
• Systematically evaluate which models perform best for specific agents, tasks, and workflows
• Evaluate which context, memory, or graph information should be available to different agents
• Implement adversarial testing, model-to-model critique, LLM-as-judge, regression testing, and systematic evaluation
• Identify failure modes and use evaluation results to continuously improve system architecture
• Optimize AI systems across quality, accuracy, reliability, latency, token usage, and cost
• Prefer measurable experimentation and evaluation over assumptions when making architectural decisions

Context, Memory, and Knowledge Systems

• Design context and memory architectures for autonomous and semi-autonomous agents
• Build with RAG, embeddings, retrieval systems, vector databases, memory systems, context graphs, and knowledge graphs
• Design how information flows between context systems and individual agents
• Determine how context should be retrieved, structured, filtered, and updated
• Build context systems that support different agents, tasks, and levels of access
• Balance context quality and completeness against token usage, latency, reliability, and cost

AI Harness Engineering

• Design, customize, and extend AI harnesses and agent development environments
• Work deeply with tools such as Claude Code, Claude Agent SDK, Codex, Cursor, MCP, agent frameworks, and similar technologies
• Build custom instructions, skills, tools, context systems, feedback loops, and evals around AI models
• Understand when existing AI tooling is sufficient and when customized harnesses or workflows are needed
• Use AI extensively throughout the engineering lifecycle while maintaining strong technical understanding, ownership, and code quality

AI Research and Experimentation

• Maintain an active, hands-on approach to staying current with rapidly evolving AI capabilities
• Regularly explore new models, research, agent architectures, frameworks, and development tools
• Turn promising research and emerging capabilities into experiments and prototypes
• Evaluate whether new approaches can meaningfully improve existing production systems
• Continuously evolve engineering practices as the AI ecosystem changes
• Demonstrate curiosity and openness to challenging existing approaches rather than relying only on established patterns

Production Engineering

• Turn ambiguous business problems into reliable production systems
• Build and maintain backend services, APIs, databases, data pipelines, and asynchronous workflows
• Contribute across the product stack when required including integrations and product-facing applications
• Deploy and operate AI systems with appropriate testing, observability, logging, and monitoring
• Design for enterprise requirements including security, permissions, data isolation, privacy, and reliability

What You Bring5 or more years of professional software engineering experience including ownership of production systems

• Deep, hands-on experience building production agentic AI and LLM systems beyond prototypes or simple API wrappers
• Strong experience designing agents, multi-agent workflows, orchestration, and tool use
• Production experience working across multiple LLMs and model providers with a strong understanding of model selection, routing, quality, reliability, latency, and cost tradeoffs
• Advanced understanding of AI evaluation systems, experimentation, and failure analysis
• Experience with adversarial evaluation, model-to-model critique, automated evals, or comparable approaches
• Strong understanding of agent, model, and context graph interactions
• Experience designing or managing how agents interact with context graphs, memory systems, knowledge graphs, or complex retrieval architectures
• Deep understanding of context engineering, RAG, memory, retrieval, and context management
• Experience with AI harness engineering or deeply customized AI development workflows
• Hands-on experience with tools such as Claude Code, Codex, Cursor, Agent SDKs, MCP, or similar
• Strong software engineering fundamentals with the ability to build reliable production systems independently
• Strong systems thinking and ability to reason about second-order effects across interconnected AI systems
• High ownership and comfort operating in ambiguous, fast-moving environments
• Strong curiosity and willingness to continuously experiment with new AI approaches
• Clear written and verbal English communication

Nice to HaveHands-on experience with knowledge graphs, context graphs, Neo4j, or other graph databases

• Experience building dynamic model-routing or agent-routing systems
• Experience building automated AI evaluation infrastructure at scale
• Experience with Postgre SQL, Redis, vector databases, and advanced search and retrieval
• Experience with Python and or Type Script
• Frontend experience with React, Next.js, or similar frameworks
• Experience with Docker, Kubernetes, cloud infrastructure, and infrastructure as code
• Experience deploying AI systems inside enterprise VPCs or private environments
• Experience with enterprise security, permissions, and sensitive data
• Founding engineer or highly autonomous early-stage engineering experience