Sayed Farhan Amjad

Worked as a data engineer for a hospital management system and designed and maintained reporting databases and power BI dashboards for hospital management. Multi-tenant databases with row and column level security both in reporting databases and interactive embedded dashboards provided our customers access to the data and the statistics that were critical to monitoring and improving the business.

Campus on Cloud is a product by Almusnet that provides a student information system for universities and colleges using the power of the Microsoft Azure Cloud platform.

I created admission data extraction modules that let the system clients request an extract of data based on a number of parameters such as admission cycle, application status, fee payment status, and more. The module was based on SQL and C# and deployed in queue-triggered Azure Function, and the status of extraction requests was maintained in a database table.

I worked on the design of the reporting database and the development of ETL pipelines in SSIS that were deployed in Azure VMs with email notification system for critical alerts.

I also worked on developing Power BI dashboards for different modules of the products, such as code, admissions, academics, and financials. We designed and embedded RLS-based multi-tenant interactive dashboards to add value for our clients.

I worked on a project to automate data extraction, feature extraction, reporting, and classification of Japanese publicly traded companies based on their yearly financial reporting and executive compensation methods.

To achieve end-to-end automation of the whole process, I first downloaded bundled reports as a ZIP file and extracted financial files with relevant data. Then, I ran feature extraction and financial information extraction for HTML files, processed and stored the information, and finally trained and tested an ensemble model to classify companies into five classes.

Previously, the client did all these steps manually, spending days processing and extracting the information. After the project was complete, the whole process took less than 10 minutes to run end-to-end and produce classification results.

Starzplay is a video streaming company with viewers based in middle-east, north Africa, and south Asia. The project was focused on understanding user behavior that leads to churn. We extracted features from the historical data of users viewing history, payment methods, language, and location data to train machine learning models based on the extracted features to identify user behavior that leads to churn from the platform. The goal was to target those users with specialized marketing campaigns to increase customer retention.