AI Engineer, Risk & Collections Agents (Claude)

Quona Capital; Ecatepec de Morelos; On-site; MXN 700,000 - 1,150,000; Full time

Job summary

Quona Capital in Mexico is seeking a skilled AI engineer to build and maintain our AI agents for Risk and Collections, leveraging Claude (Anthropic) as the foundation. You will create a reliable, parallel processing platform and enable secure, auditable access for external agencies.

The role covers agent development, cost-efficient LLM operations, robust evals, and strict access controls, collaborating with Risk Data Science to validate results and integrate feedback.

Qualifications

• Artificial Intelligence: Technical mastery of Claude and advanced Prompt Engineering; proven experience building applications with LLMs and writing evaluations (evals).
• Software Engineering: 3+ years of experience deploying Python in production and advanced SQL skills.
• Security & Access: Experience implementing authentication and authorization in web applications (OIDC / Okta or similar).
• Cloud & DevOps: Solid understanding of Git, CI/CD (GitLab CI), and AWS at a developer level, including handling sensitive data and secrets.
• Desirable: Background in Credit/Collections, experience with MySQL/Athena, and familiarity with the Claude Agent SDK.

Responsibilities

• Agent Development: Maintain and expand our current AI agents (KI Forensics) and build new Risk and Collections agents capable of publishing structured, evidence-backed findings.
• Performance & Efficiency: Implement parallel processing and optimize LLM operational costs (caching, batching, model routing) to handle a higher daily volume of cases at a lower cost.
• Reliability & Evals: Guarantee the reliability of the daily processing cycle (retries, error handling) and maintain rigorous evaluation frameworks (evals) and prompt versioning.
• Security & Access Control: Build strict access controls (OIDC/Okta) for external collection agencies, ensuring data visibility is strictly limited to assigned cases with per-user audit trails.
• Cross-functional Collaboration: Work closely with the Risk Data Science team to methodologically validate AI findings and integrate feedback from external agencies.

Skills

Prompt engineering

LLM integration

Eval design

Tools

Python

SQL

OIDC/Okta

Job description

Quona Capital in Mexico is seeking a skilled AI engineer to build and maintain our AI agents for Risk and Collections, leveraging Claude (Anthropic) as the foundation. You will create a reliable, parallel processing platform and enable secure, auditable access for external agencies.

The role covers agent development, cost-efficient LLM operations, robust evals, and strict access controls, collaborating with Risk Data Science to validate results and integrate feedback.