Ismail Affaf

– Data Pipelines Implementation from scratch the full orchestrated ELT pipelines feeding the Datawarehouse Bigquery and hoster the application on google cloud based on Kubernetes. – Data modeling Led data modeling and build various tables and dashboards to provide insights to answer key analytical questions. – Data ops Build Airflow and DBT on top of GKE and industrialize the CI/CD pipe – Technical Stack GCP, AWS, BigQuery, Airflow, DBT, SQL, Kubernetes, Spark, gitlab Data viz Looker, metabase

– Delivered full Bigquery and DBT training for different Metro department (100+ users). – Providing production best practises for airflow and DBT in data pipeline and led migration towards Databricks (Spark). – Participate in the Migration from on-premise datawarehouse (Teradata) and Control-M scheduler towards Google cloud stack based on Bigquery/Snowflake and airflow (google compose) – Technical Stack GCP, BigQuery, Snowflake, Airflow, DBT, SQL, Kubernetes, gitlab *Data viz: Looker, metabase

– Data Ops industrialize ETL/ELT (spark and dbt) data pipeline(ci/cd) and build dashboard for DW data quality on Looker Studio – Designing and implementing server-less event-driven architecture and pipeline for data ingestion from azure storage to Google storage and processing data in Snowflake load reducing 30 percent of the resource utilization vi DBT

– Deliver Data engineering solutions supporting cloud platform analytics and data insights for the business. – Led modeling data and writing data ETL/ELT pipelines to provide datasets to answer key analytical questions. – Performing exploratory data analysis and integration of new datasets via different data sources (scrapping, api...). – Technologies(Data Stack): GCP, Cloud function (serverless architecture), Biguery, Snowflake, Terraform, Airflow, DBT, Spark/Beam, Looker. Tools: Python, Scala, Kubernetes (Docker), Gitlab(CI/CD). Agile method: Scrum. *Data viz: Looker, Power BI

– Automated ETL processes, making it easier to wrangle data and reducing 20% of ETL's execution time. – Reduce critical ETL's error detection time from 20 min to 5 min using live alerting based on Prometheus. – Led efforts to migrate team apps toward and on-demand based cloud reducing infrastructure costs. – Developed a Spark NLP based hotel recommendation model trained on data from Web scraping of hotels reviews. – Technologies (Data Stack): Spark, SQL, Oozie, Hive, Sqoop, Hadoop, Qlik sense. Tools: Python, Scala, Kubernetes. Agile method: SAFe.

– Developed a deep learning model to analyze the purchasing behavior of the clients and predict their moves. – Increased the recall of the customer behavior forecast model by 6 % by utilizing attribute dependencies over time. – Technologies: TensorFlow, Python, SQL. Agile method: Scrum.