Dejan Peshevski

Over the years, I grew a passion for building a platform that enables users from different organizations to collaborate while keeping their data private.

The existing email ecosystem offers some great features. A private server, identification of the collaborating party (also from other organizations), a clear separation of the client and the server, i.e., a wide array of email clients, each able to connect to multiple accounts/email servers. It was lacking the user's profiles, sharing of data, communication (message/chat/call), and some basic BPMS for everyday interactions.

As a result, I led and architected the early development of the Affectli platform. Each of our customers hosts their Affectli instance just as they would an email server. Yet, their users can do much more. They share their data and start or participate in business processes within their organization or other organizations running Affectli.

The core platform functionalities can be extended through apps (client and server modules), packages of domain-specific BPMS, data schema, and UI forms/views. The apps keep the system modularized but also made a pathway for a marketplace ecosystem through third-party developers.

I set up a data warehouse, including a denormalized schema, aggregations, and data partitioning.

The query execution times were improved dramatically, i.e., the same servers (CPU and memory) that used to support 30,000 documents now invested 100 million.

The existing API included an entry point for submitting documents directly processed/analyzed here. I implemented a data pipeline by de-coupling the processing steps into separate tasks. The implementation improved API response times with more straightforward, faster, and queued requests to the ML (Python) module.

I designed a data warehouse for reporting, implemented a REST API in Node.js to integrate with other third-party data sources, process data, and provide data-source endpoints for the PowerBI service. I also implemented a simple dashboard for system configuration and administration.

Within Power BI, I implemented the following:
• Reports and dashboards
• Data transformations using Power Query for modeling the report's data source
• DAX expressions in the report's UI
• Row-level security limiting data scope for different user roles
• Scheduling refreshing data and export of reports

I also modeled data to support the fast query execution needed for the dashboards (materialized views, aggregation, and so on).

I designed the data warehouse used for reporting, integrated with internal organization databases and other third-party data sources using Pentaho Kettle.
I then implemented reports and interactive dashboards in JasperReports. Lastly, I implemented aggregations, materialized views, and PL/SQL functions in the data warehouse used in integrations and fast query execution for reporting.