• Energy Distribution Inc. Big Data Platform Setup
    It is a platform with servers, all of which are open source software, with batch and real-time data processing. Apache Spark, Apache Kafka, Kafka Connect, Jenkins and MinIO Object Storage.
  • Energy Distribution Inc. Big Data Platform Administration Support
  • Energy Distribution Inc. Public Lighting Fault Detection System
    The system detects faulty luminaires that do not light at night and luminaires that burn in vain during the day, and assigns work orders to field teams. The project was carried out on a big data platform consisting entirely of open source software. Apache Spark, Jenkins, Python.
  • Revenue Administration Hadoop Platform Installation and Administration
    It was carried out by Erkan ŞİRİN in his working life. Using the Hortonworks package, the Hadoop platform with 13 servers was installed and managed for one year.
  • Opet Elasticsearch Cluster Installation and Management
    It was carried out by Erkan ŞİRİN in his working life. Open source Elasticsearch cluster with 3 servers was installed and management service was provided.
  • Aisle-3 Data Platform Installation (UK)
    It was carried out by Erkan ŞİRİN in his working life. Setup of the data platform that provides Elasticsearch and ML Engineering teams consumable data. Orchestrator was Amazon Airflow, storage was S3, ETL development with Python and Pandas, database MySQL and cloud was AWS. Data source scraping data from 50 different sneakers site.
  • Complaint Classification with Machine Learning
    It was carried out by Erkan ŞİRİN in his working life. Complaints and requests coming to municipalities through various channels are assigned to the relevant units of the municipality by human operators. The ML model, which understands which unit is responsible for the complaints, is presented with an API and integrated into the complaint management software. Thus, human operator dependency is eliminated.
  • Electricity Consumption Fraud Detection with Machine Learning
    The machine learning model, which detects losses and leaks in electricity used for agricultural irrigation, works daily in a live environment with all MLOps processes.