Join CI&T in building the platform components that help teams develop, launch, and support AI agents. You’ll work across SDKs, agent frameworks, memory and context services, registry integrations, and APIs, helping turn agent capabilities into dependable tools for real-world use. CI&T works with large enterprises on AI deployment, modernization, data, marketing technology, and business strategy.

Responsibilities

• Develop Python SDK features and adapters for agent frameworks such as Strands, Claude Agent SDK, and LangGraph.
• Build capabilities for memory, skills, prompt management, and evaluation, using services such as AgentCore Memory and Redis.
• Create and maintain agent and MCP server templates for AgentCore Runtime and ECS Fargate, using HTTP and A2A.
• Integrate with the AWS Agent Registry, including registration, approval workflows, and identity-scoped discovery.
• Build and support RAG and knowledge services using Bedrock Knowledge Bases, OpenSearch vector search, and ingestion pipelines.
• Write contract and integration tests, and maintain the release parity gate.
• Help teams onboard and troubleshoot their agents, sharing learnings with the platform team.
• Participate in on-call or office hours and contribute to root cause analyses.

Required qualifications

• Advanced English proficiency (C1 or above), with the ability to communicate independently with international stakeholders.
• Extensive software engineering experience and strong Python skills, including FastAPI or a similar framework, asynchronous programming, packaging, and testing.
• At least one year of experience building LLM-powered applications or agents used by real users in a pilot or production setting.
• Experience with tool calling, prompt design, and handling model failures.
• Familiarity with at least one agent framework, such as Strands, LangGraph, Claude Agent SDK, OpenAI Agents SDK, CrewAI, or LlamaIndex.
• AWS experience with Lambda, ECS/Fargate, IAM, S3, DynamoDB, and Bedrock or SageMaker.
• RAG experience, including embeddings, vector stores, chunking, and retrieval tuning.
• Familiarity with CI/CD, Docker containers, and infrastructure as code; Terraform is preferred.

Preferred qualifications

• Experience with Bedrock AgentCore, including Runtime, Harness, Gateway, Memory, Identity, or Code Interpreter.
• MCP server development or experience with the A2A protocol.
• Experience with Redis/ElastiCache, OpenSearch, Langfuse or similar LLM observability tools, prompt management, or evaluation tools.
• Experience creating internal developer documentation and examples, such as notebooks and quick starts.
• AWS certifications or recent study or project experience in agent technology may be considered in place of multiple years of hands-on experience.

What to expect
You’ll take ownership of platform components from development through ongoing support, working closely with teams building agent use cases. Your work will help make agent development more consistent, testable, and accessible across projects.