Project Portfolio

Challenge: Difficulty with regular updates of a monolithic GLPI setup, database corruption risks on production during plugin testing, and slow interface response under load.

Solution: Containerized the high-load GLPI system using Docker. Implemented strict environment isolation between Dev, Staging, and Production. Optimized the MySQL database (added correct indexes and Redis caching) and developed custom plugins to automate internal SLAs.

Result: System update and patch application time reduced from hours to minutes (via container rebuild). Interface response speed tripled, and production data remains secure during new plugin testing.

Challenge: Lacking unified print cost accounting and control, posing high security risks for confidential document exposure on paper in the banking sector (state and commercial banks of Ukraine).

Solution: Designed and implemented an enterprise print management architecture using YSoft SafeQ 6. Configured pull printing via access cards (Secure Queue), policy-based cost saving, and automated billing reports.

Result: Delivered a unified print audit system for thousands of bank employees, reduced paper and toner costs by 25-30%, and mitigated confidential document leaks at MFP terminals.

Challenge: Manual planning of engineer visits took hours and failed to account for traffic, ticket priorities, or customer time windows, leading to delays and excessive fuel expenses.

Solution: Developed a standalone web platform and a GLPI plugin. The system automatically gathers coordinates via Google Maps API, calculates distance matrices using a local OpenRouteService instance, and passes them to the Google OR-Tools solver for optimization.

Result: Route planning time decreased from several hours to 2 minutes. Fleet mileage was reduced by 18-22%, and human errors in ticket distribution were completely eliminated.

Challenge: Handling large volumes of unstructured logs, system alerts, and incoming support emails into GLPI without proper filtering, classification, or incident pattern analysis.

Solution: Developed and deployed background Python worker services running in Docker containers. They intercept the message stream, parse and structure payloads (using regex/BeautifulSoup), write to a local database, and generate analytics metrics.

Result: Automated classification and import for 100% of unstructured message feeds. Established a flexible analytics system that exposed infrastructure weak points before they caused downtime.

Challenge

Delayed engineer response times to critical incidents due to the need to constantly monitor the GLPI web UI or email mailboxes, especially when away from desks.

Solution

Designed and launched a system of interactive Telegram bots utilizing the Telegram API and Python. Bots deliver instant critical incident notifications and allow engineers to claim tickets directly from the messenger.

Result

Initial response time (SLA) for critical incidents dropped from 15 minutes to under 1 minute. Improved mobility and continuous coverage for on-call support staff.

Challenge

Operators manually copied customer data and shipment descriptions from GLPI tickets to Nova Poshta's personal cabinet. This took 3–5 minutes per shipment and frequently led to errors in details.

Solution

Created a native PHP plugin that integrates Nova Poshta's API directly into the GLPI ticket form. Added a volumetric weight calculator, real-time branch search with debouncing, and automatic EDRPOU validation for corporate clients.

Result

Express waybill (TTN) creation time was cut from 4 minutes to 20 seconds. Shipment details errors were reduced to zero.

Challenge

Manual tracking of delivery statuses for hundreds of Nova Poshta cargo shipments containing hardware and spare parts, leading to delayed logistics ticket closures.

Solution

Developed a background Python worker service that polls Nova Poshta's API, updates delivery progress directly in GLPI tickets, and automatically triggers recipient and observer notifications.

Result

Fully automated delivery status tracking and ticket closures upon cargo reception, reducing workload on logistics staff and improving end-user visibility.

Challenge

High complexity of manually calculating contractor service costs across thousands of monthly tickets, posing billing error risks due to varying price agreements.

Solution

Developed the TicketBillingFinance plugin for GLPI. It automatically calculates closed ticket costs for billing periods based on individual contractor tariff rates and generates unified financial reports.

Result

Reduced the monthly contractor billing process from several days to 5 minutes. Eliminated human errors and ensured 100% financial transparency.

Challenge: Chaotic cartridge inventory tracking in warehouses, missing historical data for refills, restorations, and disposals, resulting in unexpected supply shortages.

Solution: Developed the CartridgeService plugin featuring dedicated Reception/Dispatch workstations. The module generates unique barcodes for each physical unit, logs refill/restoration cycles, and integrates with GLPI ticketing.

Result: Established complete control over consumables inventory in warehouses. Warehouse audit speed increased 5-fold, and unexpected printer downtime was eliminated.

Challenge: Complicated and slow cartridge allocation and consumption tracking for support engineers using default fragmented GLPI forms, leading to missing inventory records.

Solution: Developed a native GLPI plugin that embeds an interactive table directly into the ticket form. Engineers can select cartridge models, specify the service type (replacement, refill, restoration), and enter quantity in a single click.

Result: Simplified daily operations for support engineers, eliminated consumables tracking errors, and automated data collection for inventory planning.

Challenge: Collecting print counter readings from hundreds of network printers across different locations, lacking tools to calculate print costs and billing automatically.

Solution: Developed the PrinterCounter plugin for GLPI. The module retrieves printer counters automatically via GLPI Inventory (SNMP) or accepts manual inputs, maintains historical records, and calculates expenses based on page cost.

Result: Automated print counter audits across all departments, provided finance with accurate billing reports, and optimized paper and maintenance contract budgets.

Challenge: Support agents wasted critical time during initial ticket triage identifying the caller's PC specifications, software setup, or active cloud subscriptions.

Solution: Created the AssetPassport plugin to consolidate all user assets into a single "Technical Passport". The system automatically posts the user's asset layout as a private comment or description within newly created tickets.

Result: Ticket response and troubleshooting speed improved by 40% due to instant access to caller workspace configuration details.

Challenge: Processing support tickets from multiple banking clients delivering data in diverse formats (JSON, XML, email), requiring slow manual bidirectional updates of statuses and comments.

Solution: Designed and deployed an integration pipeline powered by Apache NiFi and Python. The system automatically ingests incoming requests, parses heterogeneous payloads, extracts attachments, and synchronizes comments/statuses back in each client's specific format.

Result: Fully automated data exchange between heterogeneous platforms, eliminated manual entry of comments and attachments, and reduced bank communication latency to zero.

Challenge: A high volume of errors, formatting inconsistencies, and duplicates in the client address database (20,000+ records), which prevented automated logistics, geofencing, and route planning.

Solution: Built an automated data enrichment pipeline using Apache NiFi and Google Geocoding API. The system cleans text address strings, validates them via the API, and enriches them with precise latitude/longitude, regional data, and zip codes.

Result: Successfully normalized and standardized 100% of the address database, enabling fully automated transport routing and eliminating manual verification errors.

Challenge: Exchanging tickets, statuses, comments, and attachments securely and in real time between the customer's internal system and contractors' external systems without providing direct access to the internal enterprise network.

Solution: Designed and built an asynchronous integration API gateway using FastAPI, Apache NiFi, and RabbitMQ message queues. The service translates GLPI entities bidirectionally via REST APIs, controls traffic flow rate (back-pressure), and automatically recovers transmission after connectivity failures.

Result: Automated exchange for ~1,000 tickets per month with sync latency under 5 seconds. Eliminated manual coordinator workload and ensured secure network isolation.

Challenge: The need for regular integrity verification of database backups stored as PXAR archives on PBS, and the complexity of manually restoring and migrating these datasets to Windows destination servers.

Solution: Designed a Jenkins pipeline that automatically extracts PXAR archives via proxmox-backup-client in a temporary Linux environment, copies database files via secure SCP to Windows nodes, and triggers automated database imports.

Result: Fully automated backup restoration and recovery validation in Disaster Recovery (DR) scenarios, with all configurations maintained as Infrastructure as Code (IaC) in GitHub.

Challenge: Slow and complex diagnosis of mail queues, SMTP delivery failures, and unauthorized authentication attempts in Mailcow due to fragmented Docker container logs (Postfix, Dovecot, rspamd).

Solution: Deployed a centralized log collection system utilizing Grafana Alloy and Loki. Configured automatic container discovery and enriched log entries with structured labels.

Result: Resolution time for mail issues and security anomalies cut from hours to minutes through instant error filtering in a single Grafana panel.

Challenge: SafeQ 6 relies on multiple independent TCP and HTTPS services during a single user workflow. Conventional load-balancing caused session affinity issues, leading to intermittent authentication failures and disrupted print release operations.

Solution: Designed a session-aware HAProxy architecture using shared stick tables, optimized timeout handling, health checks, and controlled failover behavior to preserve application-level session continuity across multiple service ports.

Outcome: 99.99% service availability, seamless node failover, stable operation under sustained production load, and reliable session persistence for thousands of enterprise print devices.