Mohamed Chrif Sahbeni

• • Business Intelligence Visualization: Developed 60+ Power BI dashboards dynamically pulling real-time data from Amazon RDS, and migrated them to Looker following company acquisition, showcasing key insights and KPIs across multiple business units.

• • Data Pipeline Migration: Contributed to building PySpark data pipelines for GCP migration post-acquisition, host ing data in BigQuery and enabling seamless transition from Power BI to Looker analytics ecosystem.

• • ETL and On-Demand Feature Generation: Utilized ETL processes, SQL, and pandas to generate on-demand fea tures, such as Excel files catering to the financial, marketing, and sales departments.

• • Customer Analytics Segmentation: Implemented client segmentation using rule-based and clustering algorithms, analyzing metrics like New Customer Acquisition Rate, Active Customer Rate, Promotion Redemption Rate, Cus tomer Churn Rate, and Repeat Purchase Rate.

• • Promotional Strategy Optimization: Studied promotions and deals across different regions and periods, identify ing optimal strategies and price-sensitive customer segments, resulting in 18% increase in customer retention and 37% improvement in coupon usage.

• • Marketing Process Automation: Automated distribution of personalized marketing emails to segmented audi ences and streamlined invoice delivery to restaurants through API integrations with Power Automate.

• • Holistic Student Segmentation: Segmented students based on academic performance, incorporating factors such as GPA, baseline SAT practice test scores, and target score goals. As well as Incorporating

• • Utilized Beautiful Soup and Selenium frameworks to scrape and extract needed structured and unstructured data, including text, tables, images, and links from various websites.

Subject: A conversational AI assistant in the education sector using Rasa

• • NLU (Natural Language Understanding): Understanding the incoming messages' topic using both DIET Classifier and SkLearn intent classifier, combined with SpaCy word embeddings.

• • Implementation of various functions for information extraction.

• • Dialogue Manager : decides the bot's best response based on a transformer architecture TED: Transformer Em bedding Dialogue