Dogugun Ozkaya

Amperecloud GmbH: Built data pipeline from MongoDB and VictoriaMetrics into Redis. Developed generation and loss forecasting tool in PV-energy facility. BCG Gamma: Developed a B2B machine learning project for propensity analysis. I designed and implemented an API to access model outputs and SHAP values using FastAPI. Automated the data pipeline, training, deployment, and scoring processes in a containerized cloud environment. Short Term Engagements: Implemented Data Engineering and ML Engineering projects

Designed and implemented the data pipeline from loyalty application to MapR using Pyspark. The data pipeline ingests exported files then saves them on Apache Hive. Introduced snowflake data model for the ingested data on Apache Hive. Working with the consultancy team, I created a simulation study for alternative loyalty engagement strategies. I implemented a Customer Lifetime Value prediction model and recommendation engine, to be used in customer segmentation product.

Designed and implemented the near real-time data pipeline using Apache ni-fi. The data pipeline collected data from SQL Server, Osisoft PI and government reports and saved on DWH in SQL Server. Also designed the data model. I designed and developed predictive maintenance solutions for thermal plants through time series forecasting. Additionally, I developed a model to forecast electricity Market Clearance Price with the trading team. Created near-real time dashboards at both plant and portfolio level

Following the development of online store, developed a recommendation engine with Pyspark to improve conversion. Collaborated with the e-commerce team to optimize marketing strategies using propensity models and statistical analysis. Designed and developed backend services for an in- house CRM tool, focusing on NoSQL, particularly Couchbase.

Thesis: “Comparison of Machine Learning Techniques in Blood Glucose Prediction on AIDA simulator.”. IEEE Sponsored ICIETS 2018 Conference: “Using Simulation Generated Synthetic Data for Benchmark Testing of Blood Glucose Prediction Algorithms”