Sandra Kozomara

As the team manager, I led the implementation of a new cloud-based global Data Hub for a pharmaceutical company, serving as the one source of truth for reporting across the Stada Group.

There were five main goals of the project:
1. Establish a new cloud DWH with processes around it, utilizing Snowflake, ADF, and Qlik Replicate, with CI/CD pipelines in place.
2. Migrate BI tools to the cloud, including Power BI, Tableau, and Alteryx.
3. Upskill data engineers and BI developers to use new tools and processes.
4. Migrated three legacy DWHs and dashboards to a new cloud BI environment
5. Develop BI self-service community and support businesses in the migration of their projects.

My role was to lead the project, manage stakeholders and external consultants, and organize the team. Technical lead for process and tool setup was done by data architects.

As the main developer, technical lead, and team manager, I created Location Intelligence for a telecommunication company by setting up a Location DWH and Location BI for data visualization.

Location DWH was built using PostgreSQL and ELT processes, with transformations done using SQL procedures (strong database knowledge). Data sources integrated within DWH included internal data from companies BSS/CRM, mobile data from fieldwork, British Telecom asset API, Royal Mail Postcode, Ordnance Survey address database, Census data, Competitor data, and Road work.

On top of DWH, I created Location BI to present analyzed data and KPIs such as customer density, competitor mapping, efficient route planning, and network development progress KPIs. I also used open-source technologies, such as Geoserver, QGIS, and web GIS.

I started as a one-person team and later formed a team of juniors for data visualization, maturing the team with senior database developers to maintain the growing DWH.

As the main developer, technical lead, and manager, I integrated the Oracle Spatial database with various data sources to enable maximum analytics capabilities of location data in PE Post of Serbia.

These data sources were collected from different systems, including:
• Post counter application to track productivity at the person and post office levels and monitor customer traffic to plan post office marketing activities
• Car tracking system for a vehicle usage KPI at planned working hours and misusage during unplanned working hours. Also, for KPIs regarding post-handover at exchange points.
• Postman mobile app for tracking KPIs regarding working hours, mileage, application usage, and mobile data quality.
• SAP for financial KPIs and many others.

As the project manager and full-stack developer for the Implementing Geographic Information System project, I oversaw a team of 70 people to create one of the largest GIS databases in Serbia.

The database included an address system with house numbers and geographic coordinates, making it the most comprehensive address system in the country. Additionally, it was the first system to enable geo-positioning hierarchy address elements such as municipalities, settlements, streets, PAK, and house numbers. As part of the project, we also created two additional systems for data quality and performance monitoring, which enabled the creation of new GIS postal services.

Top-level roll-out strategy tool for finding the most profitable route to lay optic cable in. The tool compared the cable length as an expense, with estimated revenue based on the count of households next to the route and the probability of signing a contract with the company.

This project involved combining expertise from the Faculty of Mathematics to provide an algorithm to search for all potential routes within 400,000 km of roads, machine learning experts to provide an ML algorithm for predicting contract signing, internal database engineers to prepare data for the two algorithms and Java developer to bring all the tools together.
My role in the project was to design a database, manage the project, and be the team lead.

The project aimed to distribute a uniform workload among parcel delivery couriers via a vehicle, especially in dense areas (tall buildings or business centers) and dispersed areas that take a long time to reach. To address this issue, I used a 1-year history of vehicle tracking data to create stopover points with two values: time needed to get to the point and time spent on the point. Based on this information, together with other information like post office location and parcel count, parcel delivery managers created 80-minute areas. This enabled us to create a uniform workload for regular (8 hours) and weekend (6 hours) shifts while preventing two different mail carriers from going to the same or similar address on the same day.
My role in this project was database engineer, providing full data analytics and visualization of a map.

To enable citizens to obtain detailed information from postal data on maps, I led the creation of a postal web GIS portal. The portal allows users to view address data and postal objects, such as post offices, cash machines, and delivery areas, which can be searched by specific parameters. My role in the project was as project lead and database developer.

The Serbian post office created a new postal address code (PAK) and needed to translate all addresses to the new PAK. My role in this project was to be the main full-stack Oracle/Delphi developer to fully support the project.

There were three main outputs of the project:
1. Create an application for collecting and maintaining a postal address system, including municipalities, settlements, streets, and PAK, in correlation to the address system.
2. Develop an application for implementing new principles in defining the postman delivery area and itinerary based on the new PAK. Also, postman productivity KPI tracking based on the new PAK.
3. Create an application for legal person address validation, enabling new service for PE Post of Serbia customers for address data validation. This application enabled holding, validating, and assigning PAK to each billing address of our customer (millions of addresses). With PAK assigned to the delivery address, customers received discounts in postal services.