Work experience

Remote (Worldwide)

• Validated and adapted 50+ open-source GitHub repositories to evaluate model behavior for frontier LLMs (Grok 4, Claude); resolved environment configuration issues across Python, SQL, MySQL, PostgreSQL, Docker, and Podman.
• Designed prompt evaluation rubrics and documented model failure modes, contributing to measurable robustness improvements in AI model performance.
• Reviewed and evaluated pull requests for correctness, test coverage, and code quality across a distributed team workflow.

San Francisco, CA, United States

• Designed and deployed a monitoring system for 5 open-source LLMs served via vLLM; implemented health probes, alerting pipelines, and real-time dashboards to ensure high availability and model reliability.
• Configured and optimized NGINX-based load balancing across multiple inference services, improving traffic distribution and reducing tail latency.
• Integrated the Datadog observability stack (metrics, logs, alerts) to monitor model performance, inference latency, and service health in production.

San Francisco, CA, United States

• Designed and implemented a global monitoring system — covering 300+ clusters and 10,000+ tasks — for a team-owned service handling 150M+ QPS; built controlled-request health probes and multi-signal alerting pipelines analyzing 10M+ QPS of health-check traffic, achieving 99.8% SLO compliance.
• Automated incident response playbooks in Python, eliminating manual remediation steps and reducing mean time to recovery (MTTR) by 30% across services handling millions of daily active users.
• Developed a multi-probe diagnostic framework to trace failures across complex distributed request paths, improving root-cause identification and reducing false-positive alert rate by 40%.
• Built and maintained dashboards and multi-signal alerts that improved real-time visibility for cross-functional infrastructure and product engineering teams.
• Participated in a global 24/7 on-call rotation spanning London and San Francisco, ensuring round-the-clock reliability for business-critical services 365 days a year.
• Partnered with infrastructure and product engineers to scale services during peak growth, balancing reliability, observability, and performance at global scale.

Vancouver, BC, Canada

• Configured and tested source code for smart warehouse automation systems deployed across multiple global locations.
• Debugged Python-based automation for package detection; verified database updates at each handling step to ensure real-time tracking accuracy.

Mexico City, CDMX, Mexico

• Enhanced ranking algorithms for job recommendation systems, optimizing user engagement and match quality at scale.
• Contributed to large-scale backend systems in Python, C++, PHP, and SQL; participated in code reviews and agile sprint cycles.