jozias tema

Designed and implemented Kubeflow and Airflow ML pipelines supporting 15+ production models and maintained feature stores and CI/CD workflows, improving model reliability and reproducibility. Implemented model monitoring and drift detection with MLflow and custom scripts, contributing to 99.5% system uptime.

Assisted deployment and monitoring of ML models across AWS/Azure/GCP and developed automated testing and validation scripts to improve pipeline robustness. Contributed to containerization and IaC practices using Docker and Terraform, streamlining cloud provisioning and deployments.

Specialized in RLHF and LLM fine-tuning, developing Python solutions for supervised fine-tuning that improved model reasoning and visualization capabilities and raised SFT performance from 55% to 75%. Conducted rigorous model testing and evaluation to enhance output accuracy across data analysis tasks.

Led analytics projects for Fortune 500 clients delivering statistical models and automated dashboards that reduced manual reporting time by 60% and improved data accuracy to 99.5%. Ran A/B tests and optimization analyses that increased conversion rates by 15% across client portfolios.

Built predictive churn and LTV models achieving 85% accuracy and created KPI dashboards serving 50+ stakeholders, enabling data-driven marketing decisions that increased retention by 12%. Analyzed large datasets (10M+ records) to identify revenue opportunities and automate reporting.

Supported market research and developed data cleaning procedures that improved processing efficiency by 40% and automated reporting pipelines reducing manual work by 50%. Produced interactive visualizations and maintained documentation for analytical best practices.

Grade: 3.1/4.0

Completed a Bachelor of Science in Applied Mathematics at the University of Yaoundé I, graduating in 2018 with a 3.1/4.0 GPA.