About
Marketing lead and content strategist specializing in AI production. I work at the intersection of prompt engineering and visual storytelling — building visual identities that stay consistent and on-character across dozens of AI generations.
For the past 1.5 years, I've led product marketing and creative direction for a Telegram AI platform with 25M+ unique users: grew its flagship channel to 1M+ subscribers, directed every video and photo creative behind it, and led the launch of its English- and Spanish-language markets. Before that: five years as Head of Marketing for a 10M+/month online education business, leading an 8-person content team and growing organic reach 200% with zero ad spend.
Two projects below show how I work as a visual artist: a five-stage character-consistency series, and a video carried through a full post-production cycle.
Experience
Product Marketing & Growth
Own product marketing and channel growth across a Telegram AI ecosystem with 25M+ unique users, including a flagship channel grown to 1M+ subscribers. Concept and direct every video and photo creative for the product. Conceived and produced the app's built-in trend template library — the tap-to-try photo and Kling-powered video effects on the home feed — end to end, from prompt design to final asset. Led creative and go-to-market strategy for the platform's expansion into English- and Spanish-speaking markets. Full-cycle AI content production — prompting, generation, editing — end to end.
Head of Marketing
Built and led an 8-person content team for a 10M+/month online school. Owned funnels, webinar and script writing, landing pages, and cross-platform analytics. Grew organic social reach 200% and launched a YouTube channel to 100K weekly views from zero — no ad spend.
Channel Creatives
A sample of concept photo and illustration creatives I directed for the channel — trend formats, seasonal content, and 3D-illustrated weekly digest covers, all built to hold attention in a fast-scrolling feed.
AI Bot #1
The Telegram bot the channel promotes — turns any selfie into on-trend AI photo and video content in a couple of taps: a constantly refreshed library of trend templates, Kling-powered video effects, and character-consistent scene generation, all inside the chat.
Character Consistency
Two original mascots, each taken through five growth stages — from cub to guardian — keeping face, style, lighting, and world consistent at every step, while each stage adds a new layer of meaning (artifact, symbol of power).
White Tiger
Cub → Guardian of Fortunes
Lynx
Kitten → Queen of the Ruins
Fashion Consistency
Worked with generative product-photo tools to turn a single reference into a full consistent shoot: same face, same body, same faux-fur fit and belt across nine separate generations — front, back, three-quarter, and close-up detail shots, ready to drop into a product page as a gallery slider.
Editing & Localization
Full post-production cycle for two AI-generated cinematic clips: seamless merge into one scene, dialogue transcription and translation, and captions styled for vertical platforms — bold type, short chunks instead of full sentences.
Close attention to timing: captions are split to be read in a fraction of a second without breaking the pace of the shot.
Original Series
Wrote the script and produced this vertical dark-fantasy/romantasy series end to end — from story beats to final cut — using Seedance, Kling, and Grok for shot and video generation, and ElevenLabs for voice, with Grok doubling as the scriptwriting and story-development tool. Same heroine, same world, same visual grade held consistent across three separate generations and scenes.
Ep. 1 — Awakening
Cold open on the heroine waking in shock, then a moonlit flashback on a cliff's edge — sets the world and the two leads.
Ep. 2 — The Dungeon
She descends into a chained underground lair and comes face to face with what's been kept there — a dragon.
Ep. 3 — The Duel
Confrontation with the shadow beast — blade drawn, backstory revealed in captions, cut to the creature's eyes in the dark.
Skills
Prompt Eng.
Iterative prompt refinement for a specific shot: angle, framing, composition, light — with control at every step.
Consistency
Keeping face, style, and world consistent across dozens of generations in a row, without the character drifting.
Multi-Model
Nano Banana 2/Pro, GPT Image 2, Kling, Veo, Seedream — picking the right model for the task, not a default one.
Video Post
Assembling generated clips into scenes, captions, timing, adapting for short vertical format.
Visual Eye
Knowing where a close-up lands, where the beat is, where to pause — always in service of the story.
Fast Iteration
Comfortable working at speed with frequent feedback and on-the-fly revisions.
18+ Scenes
Full-cycle intimate scene and cutscene generation for books and games in Seedream — staging, mood lighting, camera composition, keeping characters and setting consistent across a whole scene sequence. Samples available privately on request.