Eric Redondo

I’m a Senior AI/ML Engineer with 10+ years building production-grade machine learning and GenAI systems at scale. I specialize in LLM-powered applications, retrieval-augmented generation (RAG), and MLOps that keep models reliable in high-availability environments.

In my recent work, I designed and deployed production RAG for enterprise knowledge search serving 50K+ internal users with low-latency (<1.2s p95) responses. I built end-to-end evaluation and monitoring, including prompt regression testing in CI, LLM-as-judge scoring, retrieval confidence for hallucination mitigation, and embedding drift detection using cosine similarity distribution shift and PSI metrics.

I focus on measurable impact and operational efficiency. I reduced inference cost by 27% using dynamic context window trimming, response caching, and multi-model routing, while also cutting deployment cycles from weeks to days through CI/CD automation with MLflow model registry and Git-based workflows.

Earlier, I built scalable ML systems and data infrastructure—designing cloud-native pipelines with SageMaker, EKS, and Lambda, and developing distributed ETL and real-time streaming using Spark, Kafka, and Kinesis. I’ve worked across NLP/CV and analytics use cases, partnering with scientists to productionize models and improving performance and reliability through monitoring, drift detection, and governance standards.