About this role
An engineer who relishes taking a customer support agent from a basic demo to a robust system handling real client traffic across time zones will thrive in this position. Cobre operates an instant B2B payments platform in Latin America, solving complex financial challenges with advanced digital infrastructure. Within the AI Engineering team, the internal solutions squad builds the agentic layer that transforms customer support by resolving inquiries through real company systems and escalating complex cases to specialists.
This practitioner will own the end-to-end support architecture, migrating email automation off ad hoc flows and onto a unified decision engine spanning portal chat, WhatsApp, and Slack. The role exists to scale client support safely while enforcing strict data isolation between tenants, integrating Snowflake, Salesforce, and public APIs behind well-tested interfaces. The ideal candidate brings strong backend fundamentals, cloud-native AWS experience, and a practical view of LLM-backed features in production.
What you will do
- Own the agentic support layer end-to-end by migrating email automation off ad hoc n8n flows onto a single decision engine.
- Build deterministic scaffolding around the nondeterministic core using event-driven pipelines, idempotency, retries, and dead-letter queues.
- Enforce strict client-data isolation as a core design requirement to prevent any exposure of tenant data across channels.
- Integrate against company sources of truth such as documentation, public APIs, sandboxes, Snowflake, and Salesforce.
- Design robust escalation paths with confidence thresholds and handoff logic for Customer Success and Integrations specialists.
- Instrument systems with correlation IDs, monitoring dashboards, and meaningful alerts before building new features.
- Isolate model and vendor choices behind an anti-corruption layer so provider changes require only a single adapter modification.
What you need
- Bring five or more years of experience building and operating production backend services with direct ownership post-launch.
- Demonstrate depth in at least one backend-focused language and a willingness to work in Go given the primary company stack.
- Apply distributed systems fundamentals including asynchronous processing, event-driven patterns, idempotency, and eventual consistency.
- Develop cloud-native applications on AWS using Lambda, DynamoDB, S3, and SQS or SNS alongside containers and CI/CD pipelines.
- Apply architectural judgment through hexagonal architecture, domain-driven design, and strict separation of third-party SDKs.
- Maintain high testing discipline with table-driven tests, contract tests, and reliable strategies for non-deterministic components.
- Show working knowledge of LLM-backed features in production including prompt design, tool-calling, and retrieval fallbacks.
- Possess professional fluency in Spanish for daily collaboration and working English for technical documentation.
Nice to have
- Experience with agent-orchestration frameworks like LangGraph or similar tooling.
- Experience building or maintaining MCP servers or comparable tool-exposure layers for LLMs.
- Familiarity with defining golden datasets, eval suites, and confidence thresholds for AI workflows.
- Kafka, schema registries, and contract evolution in production.
- Payments, treasury or fintech background, especially touching client integrations or support tooling.
Skills and tech stack
Questions candidates ask
Where is the role located?
This position is based in LATAM.
What languages are required for day-to-day communication?
Professional fluency in Spanish is required as the working language day to day, alongside working English for technical documentation.
What is the primary programming language used by the engineering team?
Cobre's stack is Go-first, with Python also in the mix, and engineers should be willing to work in Go.