Deployed a CNN-based vehicle classification model as a production-ready API on AWS using Docker, achieving 82% accuracy on real-world image datasets. Designed supervised credit-risk models (XGBoost, LightGBM) over 350K+ financial transactions, improving applicant risk segmentation (ROC AUC > 0.72). Engineered scalable ELT pipelines for LatAm e-commerce analytics, identifying delivery inefficiencies and revenue concentration patterns across geographic regions. Developed NLP sentiment pipelines using TF-IDF and embeddings to automate customer review classification for digital streaming platforms.

Supported high-volume transactional systems with strict SLA and uptime requirements in regulated financial environments. Architected and maintained 10+ mission-critical banking services within COBIS CTS 3.4/4.6 ecosystems, supporting high-availability operations across IBM WebSphere and JBoss environments. Reduced API integration latency by 30% through orchestration redesign between COBIS core banking, MQ services, and third-party providers. Accelerated deployment cycles by 25% by designing CI/CD automation scripts across Azure environments. Improved onboarding efficiency by 40% through structured technical documentation, process standardization, and architectural diagrams. Ensured transactional stability in high-volume financial workflows through performance tuning and SQL optimization on Sybase and Oracle systems.