20k→250k
daily audience growth in ~6 months: onboarding, referral, retention
Product Manager · 5+ yrs · Growth · AI
Not a resume —
a living product profile.
I'm Daniil. For 5+ years I have been growing products: growth, monetization, AI workflows, Web3 and B2C/B2B SaaS. Below are cases with numbers, my method, and an AI agent that honestly answers questions about my experience and checks fit for your vacancy in a minute.
The agent answers only from a verified fact base — upload a vacancy and get an honest fit check.
Numbers instead of adjectives.
What usually takes explaining at an interview — here it is numbers upfront.
$5M→$40M
daily transaction volume: segments, activation funnel, repeat usage
20→350
paying users: paywall, pricing, renewal scenarios
42 ETH/mo
~$105k per month at the subscription model peak
China-trained product taste — an edge you cannot get from books.
I lived and worked with China for several years. I saw from the inside how products become part of everyday behavior: payments, commerce, entertainment, super-app logic, dense interfaces and instant feedback loops.
For the CIS market this is a rare lens: I read user habits faster, communicate internationally with ease and turn operational chaos into requirements, metrics and manageable change.
- Reduced shipment-control losses from 20% to 2.5%
- Cut international payment commissions by 15–20%
- Accelerated the payment cycle from 7 days to <24 hours
Cases I am ready to discuss in detail.
Each case unfolds into role, context, actions, metric and lesson — interview-ready.
Growth
Scaled DAU 20k → 250k
Details Collapse- Role
- Product / Growth PM
- Context
- A Web3 product with friction points in onboarding and repeat usage.
- Actions
- Referral mechanics, gamification, retention scenarios and user segment analysis.
- Metric
- DAU grew from 20,000 to 250,000 in about 6 months.
- Lesson
- Growth works best when user motivation is built into the product's core loop.
Monetization
Grew volume $5M → $40M daily
Details Collapse- Role
- Product / Growth PM
- Context
- A high-load transaction platform with a critical activation and repeat-usage funnel.
- Actions
- Funnel/cohort analysis, A/B hypotheses, friction point hunting and repeat activity work.
- Metric
- Daily transaction volume grew from ~$5M to ~$40M.
- Lesson
- Not all users are equally valuable: segmentation grows quality activity, not just traffic.
Subscription
Grew monetization: 20 → 350 paid users
Details Collapse- Role
- Product / Growth PM / Co-founder
- Context
- Web3 automation tools with community-driven demand and subscription-based access.
- Actions
- Paywall logic, pricing, onboarding, renewal scenarios and limits for non-paying users.
- Metric
- Paid users grew from 20 to ~350; peak net revenue — ~42 ETH/mo.
- Lesson
- Monetization is not just price — it is clear value packaging and the right paywall moment.
More: an NFT launchpad MVP from scratch in 3 weeks and a GameFi Telegram Mini App with progression and battle pass — ask the agent for details.
Four reasons to invite me to an interview.
Growth without magic
I break growth into funnels, segments, activation, repeat usage, retention and paid conversion — and tie hypotheses to metrics, not feelings.
AI as an operations layer
AI agents for discovery, research, VoC, competitive analysis and requirements. CustDev analysis: 2–3 hours → 30–45 minutes. Weekly reporting: half a day → 40–60 minutes.
Monetization
Pricing, paywalls, subscriptions, paid conversion, renewals and paying-user retention. I look at monetization through LTV, ARPPU, churn and unit economics.
China & B2B context
Negotiations with suppliers in China and the UAE, international payments, operational processes — I turn chaos into requirements and metrics.
Strong fit
- Product / Growth PM — growth, activation, retention, roadmap, business metrics
- AI Product / Product Ops — agent systems, RAG/MCP, research and reporting automation
- Web3 / Crypto / FinTech PM — transactions, fees, marketplaces, subscriptions, volume
- B2C / B2B SaaS PM — discovery, delivery, analytics, stakeholders, monetization
Honestly: weaker fit
Roles that need only project management without ownership of metrics, roadmap or product decisions.
How I work.
I don't start with features. I start with a goal, a segment, a metric and a constraint — then hypotheses, prioritization, delivery and measured impact.
- 01Goalwhat should change in the business
- 02Segmentwho we are solving for
- 03Metrichow we know it got better
- 04Hypothesiswhat we test and why
- 05Deliveryfast launch without quality loss
- 06Impactmeasure and start the next iteration
What you get in the first 30 days
01 · 审Audit
Funnel, segments, churn points, current analytics and user feedback.
02 · 焦Focus
Hypotheses, impact/effort, quick wins and team-aligned priorities.
03 · 行Delivery
First production changes, metric impact check and the next roadmap iteration.
FreyrAI · AI video generation platform
From a one-line idea
to a finished video.
Script, frames, motion, voice, subtitles and final cut — one system instead of five subscriptions and a folder of loose clips. Built for creators who ship series, not one-off posts.
- End-to-endidea → final MP4
- Series-nativeconsistent cast & style
- Pre-betapipeline working today
Pre-beta · working pipeline · raising
Making AI video is still a manual, expensive relay race.
Three problems keep serious creators from scaling output.
Tool sprawl
One video means juggling a script model, an image model, a video model, a voice tool and an editor — copying assets between five tabs, by hand, every time.
Cost per clip
Platforms that resell model API credits pass the cost straight through. At production volume the unit economics stop working for the creator and the platform alike.
Consistency collapses
A character drifts between shots and breaks entirely between episodes. That kills the one format that actually compounds an audience: the series.
One brief in. One finished video out.
Five stages run as a single orchestrated pipeline — automatic end to end, or step-by-step with approval gates.
-
Script
The brief becomes a structured plan: scenes, cast, dialogue, shot specs and pacing.
-
Frames
Every scene is composed as an exact first frame, locked to the project's cast and visual style.
-
Motion
Frames are animated with motion described separately from composition, so movement never rewrites the shot.
-
Post
Upscale, voice or native sound design, subtitles and per-scene regeneration where a take missed.
-
Assemble
Transitions, title card, cover and a final MP4 that passes automated quality checks before you ever see it.
Stages pipeline into each other: frame n+1 renders while clip n is still animating. Nothing waits for the whole batch.
Why this wins.
Four advantages, in the order they matter to an investor.
Not another API wrapper.
Competitors resell model API credits at cost plus margin, so their gross margin is capped by someone else's price list. FreyrAI runs the same pipeline through a proprietary provider-abstraction layer at an order-of-magnitude lower cost per generation. The adapter design also means heavy stages can migrate to official APIs as those prices fall — without the product changing. Architecture details under NDA.
Consistency is the moat.
Each project carries an asset bible: cast appearance, style, reference frames and prior episodes. New episodes inherit all of it, so the same character survives across shots, lighting and episodes. Anyone can generate one good clip; making episode seven match episode one is what turns a tool into a habit.
The stack goes deeper than "prompt in, clip out".
A prompt compiler turns a brief into precise, model-specific instructions; storyboard gates catch a bad scene before it becomes an expensive render; every shot separates composition from motion. This is years of accumulated craft, not a wrapper a weekend hackathon reproduces.
Quality control is built in, not hoped for.
Every final render is checked automatically for container, codecs, resolution, loudness, silent tails, black frames and frozen shots — and repaired where possible. A fixed regression suite and per-run telemetry track how often the pipeline finishes with zero human intervention. The target is above 92%.
One engine, four products.
The same pipeline serves formats that would otherwise need separate tools.
Short-form series
9:16 episodes with a recurring cast, hook frames and covers.
Long-form stories
16:9 narrative video with title cards and chaptered pacing.
UGC-style ads
Photoreal creator-style spots for products, in volume variations.
AI montage
Bring your own footage: automatic edit, subtitles and revisions by chat.
Who pays for this.
All four segments share one trait: they need volume, repeatedly, on a schedule.
Short-form creators
Daily episodic content for TikTok, Reels and Shorts — one universe, a new episode every day.
YouTube storytellers
Longer narrative formats with a recurring host or character and a consistent visual world.
SMM agencies
Content for many clients in parallel: brand characters, style templates, predictable turnaround.
Performance advertisers
Fast creative variations for paid social, where testing volume decides the campaign.
Business model.
Subscription plus usage credits — the category standard, with a cost base the category does not have.
Subscription + credits
Tiers bundle a monthly credit pool, output quality and how many jobs run in parallel. Credits decouple pricing from provider cost, so operation prices can be retuned without repricing plans.
Free tier as a channel
The free tier is generous enough to finish one real video, watermarked. Activation matters more than saving on free users, and every watermarked video is distribution.
Expansion revenue
Turning one long video into a pack of shorts, agency seats and — once the platform is stable — an API. Average contract value grows without new acquisition.
Final price points are deliberately not published yet: they are being calibrated against real cost telemetry from the closed beta rather than guessed in advance.
Where it actually stands.
An honest read, because the gap between demo and product is where most of this category lives.
Working today
- Full pipeline, idea to final MP4
- Cast and style persistence across episodes
- Voice, subtitles, upscaling and assembly
- AI montage module for existing footage
- Automated quality checks and a regression suite
In progress
- Multi-user backend: accounts and persistent storage
- Job queue and worker separation for concurrency
- Billing, credit ledger and admin tooling
Next
- Closed beta with 30–50 series creators
- Cost telemetry, then public pricing
- Public launch and referral loops
Who is building it.
Daniil Lebedev — product manager with 5+ years in growth, monetization and AI workflows, and the sole builder of FreyrAI so far: product, architecture and the entire generation pipeline. Previous work includes scaling a product from 20k to 250k DAU and growing a subscription product from 20 to ~350 paying users.
Raising to turn a working pipeline into a platform.
The technology is proven end to end. The capital goes to multi-user infrastructure, closed beta and the first go-to-market motion. Happy to walk through the architecture, economics and roadmap in detail under NDA.
AI agent: ask about my experience.
The agent answers only from a verified fact base — no inventions. Ask a screening-style question or upload a vacancy: it will honestly estimate the match and show where the fit is strong and where the gaps are.
- 1Ask a question — like on a screening call
- 2Or upload a vacancy file
- 3Get an honest fit check with a match percentage
Daniil's agent
answers from a fact base · RU / EN
online
Next step
Looks like a fit? Let's check.
Message me on Telegram — I reply within a day. Or start with the resume and the agent.