2 days ago
Remote, BrazilSenior
Responsibilities
- Build production AI agents and TypeScript services on AWS using Lambda, Step Functions, DynamoDB, Bedrock, Strands or LangGraph, and AI SDKs.
- Design agent behavior, including model investigation boundaries, enforced code behavior, scoped tool access, sandboxing, budgets, and timeouts.
- Build evaluation systems using golden sets, production replays, calibrated judges, recall, precision, and false-positive-rate gates.
- Implement observability through reason-coded metrics, structured logs, alerts, Splunk, and VictoriaMetrics, and investigate production failures.
- Maintain Terraform infrastructure across environments and improve GitLab CI/CD for build, test, deployment, and quality gates.
- Build resilient integrations handling retries, rate limits, permanent failures, and stale state.
- Review AI-generated code for security, correctness, and maintainability, and help teams configure assistants, skills, hooks, and MCP servers.
- Build React-based agent-facing tools, APIs, summaries, dashboards, and internal tooling, and document systems with diagrams, runbooks, and decision records.
Requirements
- Strong production experience with TypeScript and Node, including asynchronous code, testing, and monorepo tooling.
- Production experience shipping LLM API features involving prompts, tool calling, structured output, and long context, with the ability to explain evaluations, failures, and fixes.
- Proficiency with Python for evaluations, analysis, and scripting.
- Experience with AWS serverless development, including Lambda, Step Functions, DynamoDB, IAM, event-driven design, and Bedrock.
- Ability to safely read and modify Terraform across multiple environments.
- Hands-on NoSQL and SQL experience, including live schema changes and migrations.
- Experience designing metrics, logs, and alarms for nondeterministic systems and debugging production issues.
- Proficiency with CI/CD, ideally GitLab CI/CD, and daily use and verification of AI coding assistants.
- Fluent spoken and written English with the ability to explain trade-offs and drive team adoption.
- Desirable experience with multi-agent systems, MCP servers, evaluation frameworks, model routing, prompt caching, token budgets, prompt-injection defense, agent security, React, GraphQL, Java, Spring Boot, code intelligence, retrieval, Kubernetes, AWS AppSync, EventBridge, Playwright, Cucumber, or Claude Certified Developer or Architect certification.
Benefits
- Company-paid medical insurance, mental health programs, and five undocumented sick-leave days per year.
- Internal meetups, conferences, workshops, Udemy access, language courses, and company-paid certifications.
- Long-term employment with 20 working days of paid vacation and local bank holidays.
- 100% remote work in Brazil.
- Internal Mobility Program, global projects, community growth opportunities, and regular team-building events.
Tech Stack
Amazon DynamoDBAWSCucumberGitLab CI/CDGraphQLJavaKubernetesNode.jsPythonReactSplunkSpring BootSQLTerraformTypeScript
Categories
About Ciklum
Ciklum is an IT services firm that builds custom digital products and provides nearshore engineering teams, QA, DevOps, data analytics, and UX for enterprises and high-growth startups. It operates a services model spanning product engineering, e-commerce solutions, and managed delivery centers. Founded in 2002 and headquartered in London, the privately held company serves global clients including Just Eat and Zurich Insurance.