Ikram Battal

Design and deployment of annotation and evaluation pipelines for training AI models (projects with Anthropic, Plaid, Shopify). Development and maintenance of analytical dashboards and automated reports using Databricks and Tableau to monitor KPIs, data quality, and operational performance. Audit of model outputs using LLMs (GPT-5, Claude, Gemini): error analysis, comparative evaluation, and generation of advanced metrics.

Design of an end-to-end modeling pipeline (Python, Scikit-learn) to forecast short- and medium-term demand across multiple vehicle segments. Development of automated data ingestion and preprocessing workflows using Airflow, integrating sales, production, and order data. Creation of dynamic dashboards in Power BI to visualize forecast accuracy, production gaps, and inventory coverage by segment. Deployment of an internal API using Flask to provide up-to-date forecasts to Supply Chain and Production teams.

Implementation of a real-time Big Data processing pipeline using Spark Streaming and Kafka for data ingestion, transformation, and consolidation of customer data. Design of an intelligent fraud detection system using Python, Scikit-learn, TensorFlow, and PyTorch, capable of automatically identifying anomalies in data. Advanced time series analysis of consumption data using ARIMA and LSTM architectures to model usage patterns, anticipate variations, and detect fraudulent sequences.