AI Engineer

This is a fully remote position, open to applicants in Mexico.

📋 Description

• Implement agents into production to assist members via WhatsApp and future channels.

• Coordinate agents and conduct tool calling against the company's systems.

• Build member context from multiple sources.

• Define and execute evaluations on real cases and acceptance criteria before launching.

• Create an observability platform for LLMs with traces, costs, latency, and quality metrics.

• Develop evaluations as infrastructure, ensuring access to provider-agnostic models and prompt versioning.

• Implement guardrails, PII protection, identity verification, and a human-in-the-loop approach in a regulated environment.

• Manage projects from inception to production, addressing the problem definition through to implementation.

• Document decisions using ADRs, participate in reviews and pairing sessions, and establish reusable patterns.

• Collaborate with product, data, operations, and other engineers.

• Decide on whether to build, buy, or assemble resources for LLM development.

⛳️ Requirements

• Approximately 5+ years of experience.

• Experience in deploying LLMs in production, having operated at least one system in front of real users post-launch.

• Significant experience with tool calling: designing tools and managing incorrect calls or erroneous arguments.

• Extensive experience with evaluations: real case datasets, LLM-as-judge calibrated against humans, and regression testing prior to any prompt or model changes.

• Experience handling sensitive data in production: PII, retention, and processing agreements with vendors.

• Native or fluent Spanish speaker.

• Professional English proficiency.

• Expertise in context engineering: compaction, memory, and context provisioning via tools.

• Knowledge of agent retrieval and MCP servers.

• Understanding of agent security: direct and indirect prompt injection, minimum permissions per tool, and red teaming.

• Experience with multi-agent systems, subagents, and durable execution.

• Familiarity with prompt caching, model routing, and structured outputs.

• Online evaluations: shadow mode, A/B testing between prompts or models, and human feedback returning to the dataset.

• Experience with WhatsApp, voice, health, or insurance is a plus.

🏝️ Benefits

• Feedback at every stage of the selection process.

• Default self-merge.

• Continuous integration, types, and tests as blockers.

• Architecture review focused on ADRs.

• A selection process aimed at closure within a few weeks.