Portfolio

What It Does:

• Analyzes all distribution platforms as a cohesive product ecosystem rather than siloed channels, revealing true content performance and audience behavior patterns.

Why It Matters:

Traditional analytics treat each platform separately. This dashboard connects the dots, showing how content performs across the entire media landscape.

Key Features:

• →Multi-platform content aggregation with smart filtering
• →AI-driven content creation for web, social media, newsletters, and video scripts
• →Vector search-powered reader chat interface for archive exploration
• →Direct GA4 API integration for real-time analytics
• →Smart "Read More" recommendations using vectorized database

Tech Stack:

Custom AI-powered backend, Vector search, GA4 API, Custom frontend

Impact:

Streamlined content operations and enabled new reader engagement channels through conversational interfaces.

Implementations:

Partly implemented: Startupt.in, AI For Newsroom

Results:

Data-driven content creation that continuously improves based on audience response. Enables editorial teams to scale social media presence without sacrificing quality or brand consistency.

What It Does:

• Generates social media posts across all major platforms
• Creates Telegram posts following editorial formats
• Produces interactive quizzes from any URL or article text
• Transforms raw data (Excel files) into formatted content like event listings (Smart Afisha)
• Parses URLs to extract and repurpose content
• Offers customizable output formats, styles, and tone of voice
• All prompts are fully tunable for editorial control

Intelligence Layer:

• →Fine-tunes prompts based on actual performance data
• →Learns from previous month's engagement metrics
• →Adapts to what resonates with specific audiences
• →Maintains adherence to editorial style guides

Implementations:

JAM News, Delfi.lt, Chayka.lv (partly)

Core Technology:

• →Complete database vectorization for semantic search
• →Vector search as the foundation for accurate content discovery
• →Context-aware recommendations and content relationships

Applications:

• →AI-Powered "Read More": Intelligent content recommendations based on semantic similarity rather than simple keyword matching
• →Archive Chat Interface: Conversational access to entire content archive, allowing readers to ask questions and discover content naturally
• →Enhanced Search: Goes beyond keyword matching to understand intent and context

Why Vector Search:

Most accurate and suitable method for AI-powered database interactions. Enables discovery of relevant content even when exact keywords don't match.

Implementations:

Dev.by

The Challenge:

Newsrooms need to track multiple sources, identify trending topics, and quickly repurpose content across platforms while avoiding bot restrictions.

The Solution:

• →Advanced RSS and HTML parsing with bot restriction bypassing
• →Multi-source monitoring including Google News and Telegram channels
• →AI-powered content summarization and platform-specific repurposing
• →Automated digest creation and distribution via email and Telegram

Results:

Reduced content curation time by 60% while expanding coverage scope.

Automation:

• →Scrapes regional court websites
• →Filters cases by specific legal articles
• →Creates curated lists for newsroom analysis

Impact:

Immediate, up-to-date case information for journalists

Content:

• →Real-world usage examples
• →Resources and tools
• →Industry news and developments

Tech:

Custom AI-powered CMS with bespoke frontend design.

System:

• →Custom AI-powered quiz generation and management CMS
• →Responsive frontend for engaging user experience
• →Automated daily content pipeline

Key Features:

• →Content discovery and extraction (70% automated)
• →Story repurposing and creation (70% automated)
• →Multi-platform distribution (70% automated)

Infrastructure:

Custom AI-powered CMS with integrated frontend.