Eugene Balkind

I worked as a data scientist for a client's advertising analysis project. I collected data from various sources, such as the clients' CSVs, SQL databases, and public data. I formatted the data to single time series standards and conducted extensive data analysis to identify potential lag, adstock (carry-over effect), and diminishing returns.

To build the linear regression model, I performed feature engineering, hyperparameters tuning, and lag and adstock adjustments to ensure that the model accurately predicted the client's ROI. Once the model worked, I used it to answer clients' questions about ROI and provided them with actionable insights.

I regularly updated the model with new data to provide valuable long-term insights to the client. Through this project, I demonstrated my expertise in data analysis and statistical modeling and my ability to apply this knowledge to real-world business problems.

I developed an algorithm that analyzed cryptocurrency streams from a crypto exchange (Binance), formatted the data, and suggested an optimal trading (arbitrage) strategy. The algorithm focused on BTC, ETH, altcoins, and USDT.

My responsibilities included collecting and formatting the data from various cryptocurrency streams to ensure the data was compatible with the algorithm. I then conducted extensive data analysis to identify trends and patterns in the data and used this information to suggest optimal trading strategies.

The algorithm was designed to identify arbitrage opportunities between different cryptocurrencies, including BTC (or ETH), altcoins, and USDT.

In addition to the aforementioned algorithm that analyzed cryptocurrency streams, I used LSTM to predict future rates of cryptocurrencies. By incorporating LSTM into the algorithm, I created a more sophisticated model that could make more accurate predictions based on historical data.

The LSTM model was trained on historical cryptocurrency data, allowing it to learn patterns and trends in the data. This information was then used to predict the future values of the cryptocurrencies, allowing for more informed trading decisions.

As a data scientist, I developed a cutting-edge recommendation system based on clustering that suggested projects to existing clients. My responsibilities included collecting and formatting client data, performing feature engineering, and building a clustering model and recommendation system.

To begin the project, I collected and formatted client data to ensure compatibility with the recommendation system. I then conducted extensive feature engineering to identify key features that could be used in the clustering model.

Using the identified features, I built a clustering model capable of accurately identifying and grouping clients based on their needs and preferences. Once the clustering model was working, I suggested recommended projects to the existing clients based on the needs and preferences of similar clients in the cluster.

I created a powerful script that scraped major UK job boards and filtered for suitable data science contracts. The script was written in Python using the Selenium library, allowing efficient and automated web scraping. The script was designed to scrape job boards like Indeed and TotalJobs and filter for data science contracts matching specific criteria.

I incorporated NLP techniques to improve skills matching to further enhance the script's accuracy. By analyzing the job descriptions and identifying keywords related to data science skills, the script was able to identify suitable job postings that matched the skills and requirements of the client.

Once the suitable jobs were identified, they were added to the database for future analysis. This allowed for easier tracking of suitable job postings and ensured clients were quickly informed of potential job opportunities.

I created a powerful app that allowed users to find all possible connections from one postcode to another. This included flights, trains, buses, and intercity connections, making it a comprehensive and valuable tool for travelers.

I collected and processed data from various sources, including APIs, Amadeus API, and web scraping using Selenium. This allowed for a wide range of transportation options in the app.

Although the app was initially developed as a prototype, there is potential to expand it and make it available to a broader audience. This would require further development and data collection, but the initial prototype provides a solid foundation for future work in this area.