At Grupo EFE, I engineered and operated cloud platform infrastructure for analytical environments across development, testing, and production. I standardized Docker runtimes and orchestrated workloads on Kubernetes (EKS), with auto-scaling and GPU node allocation for training.
I deployed MLflow as a centralized model registry for artifact versioning, experiment tracking, metadata, and lineage. I also implemented automated monitoring for model health, drift, latency, and business KPIs.
At Konecta, I configured asynchronous Amazon SageMaker endpoints and FastAPI/REST layers for production LLM inference. I fine-tuned Llama 3.0 with QLoRA and built RAG agents using AWS Bedrock; my automation and infrastructure-as-code work reduced model deployment time by 40%.
Earlier, I developed predictive maintenance models for vehicle fleets at Autonort Trujillo SAC, where the implemented strategies reduced unplanned downtime by 15% and increased operational availability by 20%. I also developed predictive models for clustering suicide cases in Peru using data provided by SINADEF.

