Experience

Where I've worked.

One production role, four internships each one shipping something that had to work, not just demo well.

01 — Jul 2025 – Aug 2026

Data Scientist I — Generative AI

Ominimo Insurance · Remote, Netherlands

  • Customer-facing AI assistant: built and iterated on a production GenAI assistant handling ~8,000 customer interactions per month, achieving an 80% end-to-end resolution rate with 83% positive customer feedback.
  • Automated email handling: built an AI-powered email automation system processing ~2,000 customer emails per month at 70% end-to-end resolution, with responses generated in ~4 seconds and 95% resolved inside the 24-hour SLA.
  • GenAI observability: designed and maintained the observability dashboard tracking automation rate, intent accuracy, escalation rate, resolution and response time in production.

02 — Feb 2025 – May 2025

Data Science Intern

4D Software · On-site, Morocco

  • Developed a GenAI-powered chatbot answering clients' technical issues, improving support efficiency and reducing the need for human intervention.
  • Designed a RAG pipeline with an agent and integrated Weave to evaluate model performance in production, achieving 98% accuracy in context-based responses.

03 — Jul 2024 – Oct 2024

Data Science Intern

3D Smart Factory · Remote, Morocco

  • Developed an intelligent recruitment system using LLMs and ML algorithms to match candidates with job descriptions, improving recruitment efficiency by 35% and reducing hiring timelines by 50%.

04 — Jul 2024 – Oct 2024

Data Engineering Intern

BroccoliData · Remote, Morocco

  • Analysed food dependency and food security indicators in Morocco within a containerised development environment.
  • Used dbt to transform data in DuckDB, enhancing data management processes and enabling actionable insights for project stakeholders.

05 — Jul 2023 – Sep 2023

Data Engineering Intern

Department of Computer Science, ENSAH · Remote, Morocco

  • Developed a document analysis and storage solution using HDFS and HBase, achieving a 30% improvement in data retrieval, 40% faster queries and 25% faster metadata processing.