Mohammed Ferjani

Vattenfall is a leading European energy company aiming to make fossil-free living possible within one generation. I took part in several, from scratch, product projects that share a data platform and contribute to facilitating data driven decision making, optimizing production and reducing maintenance costs. My responsibilities were :

• Gathering requirements specification in collaboration with the different stakeholders and translating them into solution proposals.

• Providing a detailed design description document of possible solutions in compliance with the existing infrastructure and security guidelines.

• Developing POCs, maturing them to MVPs and taking them to the next levels, all in a scrum agile way of working.

• Migrating from a Data Warehouse architecture to a medallion Data Lake architecture, single source of truth, to support BI, reporting, advanced analytics and data exploration needs.

• Designing and building batch and streaming based Data pipelines to fuel a multitude of data products initiatives.

• Designing and building a knowledge base system to store, manage and expose Assets' master data and metadata.

• Building tools for continuous monitoring and quality control of all data being delivered. Technical stack : Hybrid setup including Azure cloud and on premises data centers. Kafka (confluent & OSS), ADLS, Databricks, Azure Data Factory, Django Rest Framework, PostgreSQL, Timescale DB, Docker, Kubernetes, OpenSearch (ELK stack), GitHub, Artifactory, Azure pipelines (CI/CD), Azure devops, …

1 year, 10 months

ENGIE group offers energy supply, management and services including design, installation of equipment and maintenance. I took part in several IOT projects leveraging machine learning to provide predictive maintenance for : 1. wind farms (renewable energy), 2. smart buildings' energy management (Ohio State University) and 3. AHU (Air handling units). I've been tasked with designing and developing :

• Data pipelines for data ingestion, transformation and storage : Real-time mode using Kinesis Data Streams and Batch mode using S3 buckets or aws Batch both triggering lambda functions (that scales out/in automatically), SQS and Influx DB hosted on EC2.

• REST services to use (retrieve and aggregate) both data coming from the sensors as well as ML (predictions, recommendations and alerting).

• Preventive monitoring to supervise all workflows end to end, from data acquisition to machine learning outputs. Technical stack : AWS cloud (S3, Kinesis Data Streams, Lambda functions, EC2, RDS (PostgreSQL), SQS, SNS), python (REST services with Flask framework), SQLAlchemy, Marshmallow, PostgreSQL, timeseries InfluxDB, Sentry, Docker, Jenkins, Artifactory, C3IOT platform, Grafana, Splunk, GitHub, Jira, Confluence, …

15 months

Photobox is an european market-leader in creating and personalizing photo-based products. I've been in charge of :

• Developing mico-services to manage the storage of all customers' creations and check their conformity before the orders go into production process.

• Processing massive amounts of data, using Hadoop tools in order to produce sales reporting in batch mode (sqoop jobs, Oozie, HDFS, pyspark jobs, hive) or in real time by processing streams of data (flume, apache Kafka, spark streaming, hive) Technical stack : REST services with Jersey (Java/JEE), Spring (mvc, security, data), Tomcat, AWS services : NoSQL DynamoDb, ElasticSearch, Serverless Lambda applications, DataPipelines, Dead-Letter Queues, Kinesis, etc. Git, Maven, Nexus, Hudson, Ansible, Gatling, Jira, Confluence, Hadoop : HDFS, sqoop, Hive, pyspark.