David legarre saavedra

Helped build an AI assistant for banking teams, collaborating with product and domain experts to improve support workflows. Collaborated with product owners, domain experts, and business stakeholders to translate operational requirements into production-ready AI features.

• Developed and deployed LLM-powered chatbots for WhatsApp and voice channels using FastAPI, LangChain, and LangSmith, achieving ˜70% automated customer support resolution in production. • Containerized ML/LLM microservices with Docker and deployed them on an on-premise Kubernetes cluster, configuring autoscaling, health checks, horizontal pod autoscaling, and secure secret management. • Built CI/CD pipelines to automate building, testing, and deployment of backend services, ML models, and APIs, improving d

• Built predictive models using wearable sensor data to forecast optimal medication timing for patients, achieving 93% accuracy and supporting adherence improvement. • Developed interactive dashboards in Power BI and visualizations in Python/R to surface trends, anomalies, and KPIs for clinical stakeholders. • Performed data cleaning, feature engineering, metric definition, and model validation using statistical methods. • Automated repetitive reporting tasks to reduce manual workloads and improve consisten

• Developed computer vision models to detect cardiac structures in MRI scans, automating parts of diagnostic workflows. Preprocessed medical imaging data, corrected artifacts, and performed benchmarking of model architectures.