Data Scientist

7/2021 – 8/2021

- A town wants to live on 100% renewable energy. It has enough solar panels and wind turbines to generate electricity, so they disconnected from the main electricity grid a while ago. However, they usually do not use the electricity at the moment of its generation by sun and/or wind, which is causing daily power blackouts. They are fed up with all the electricity problems and asks help to adjust the electricity demand to the electricity generation.

- We took an iterative approach to have an end-to-end system from the first iteration.

- We trained the model using simple naive forecast: average consumption of the same day of the week, considering N weeks before further iterations.
- For Model evaluation: we backtest with a sliding window approach, optimized for MAPE
- Model deployment: deploy the best performing model as an Azure Machine Learning Batch pipeline Model

- Model post-processing:
- Compare energy consumption forecast with energy generation from wind and solar predictions.
- Classify the energy consumption per day (normal, middle, low charge), using consumption distrubtion pattern persist results to Azure Blob Storage to be used by Power BI.
- Finally, build dashboard with recommendations for the energy consumption on the following day.

Microsoft Azure, Microsoft Power BI, Python, Scikit-learn

2/2021 – 5/2021

- Perform data-driven insights across a wide range of topics and business industry domains by developing machine learning algorithms.
- Project 1 - Google Store Analytics: Market and customer analysis of Google merchandise store to predict customer churn, shopping behavior, and sales revenue ver time using machine learning models.
- Project 2 - Kickstarter Project: Presenting our stakeholder to help find out if the startup or project will be able to get funds on Kickstarter platform using ML

Amazon Web Services (AWS), Data Science, Deeplearning4j, Machinelles Lernen (allg.), Microsoft Azure, Microsoft Power BI, Pandas Datenrahmen, Python, Scikit-learn, TensorFlow

12/2017 – 10/2020

• Characterization of laser diodes using streak camera.
• Analyzing the dynamic behavior of lasers using mathematical simulation models.
• Conducted practical lessons on “Internet of Things” (IoT) using microcontroller esp32 to build heart rate monitor that takes measurement using Bluetooth low-energy (BLE) and deploying it on app to send and receive data in real-time.
• Fourier analysis and transformation of signals and studying the data.
• Digital Signal Processing (FFT) library used to convert the signal.

C++, Internet of Things (IoT), Python