Operating an unlicensed AI product pipeline
Product Manager — ALCONOST — localization company, ~100 people · committed remotely
AI playable-ads pipeline — alconost.video/playable-ads
Built an AI-driven production pipeline that helps app and game companies scale user acquisition. Auto-generates playable HTML5 game ads plus matching video cuts, with every platform resize (Meta, Google, TikTok, Unity, AppLovin) produced automatically. Concepts are prototyped, polished into network-ready ads, then hyperlocalized into hundreds of audience-specific variants — more ads delivered, more ideas tested, more conversions for clients.
AI Data Labeling Platform — multimedialabeling.com (0 → live)
Built a full-stack platform that streamlines AI training-data collection and QA: multi-language audio recording, speaker qualification forms, automated quality scoring, and an admin dashboard. Stack: Lovable, Supabase, Cursor. Result: easy, manageable, continuous delivery of high-quality datasets.
Alconost MT suite — alconost.mt (evaluate, glossary)
Owns product management for the MT suite. Shipped the core app that evaluates translations across multiple LLMs, plus a glossary tool. Aligned with CEO and CMO, ran design workshops, managed engineers, shipped through release, then improved via A/B tests, fake-door experiments, and analytics.
Kraken — internal marketing intelligence tool (0 → live)
Identified manual app-store data collection as the team's biggest bottleneck. Built a small team (one designer, two engineers) and shipped a web app aggregating data from App Store, Google Play, Steam, Nintendo, and Epic Games. Now powers 3+ ABM campaigns a month.
alconost.com redesign + multilingual rollout
Used GA4 and Hotjar session replays to find funnel leaks, coordinated an international team on regional design and content, managed localization in Crowdin with AI translation + human post-editing. Result: lifted conversions across target regions.