Product& Growth
I build growth into a single instrumented machine and own every part of it, from the strategy down to the code.
Measurement goes in first, and every downstream stage answers to what it records.
Shipped for:KuCoinEntangleArtradeQuraniumBitazzaOlmiSpree
Growth stack
I build the whole growth machine inside a company: acquisition, activation, lifecycle, owned channels, pricing, testing. Every piece here is something I've shipped for a real product.
Every action a user takes becomes a number, and every number I report traces back to the real event behind it. When I tell you activation moved, you can check me.
I map every route a stranger takes to the product and tag it at the source, so paid, referral, partner content, and events all land in one attribution map. When a user shows up, I already know which channel produced them.
I build the loop that turns a signup into a returning user: identity binding, a verified action, XP settlement, a burn or redemption, then repeat. Every stage is gated and instrumented, so I can see where a cohort stalls and re-trigger it.
I pitched, designed, and built Swyftx Learn from zero, and that work informed the KuCoin Learn approach that came after.
I build the messaging that brings users back: lifecycle sequences, a CRM, and re-engagement for cohorts that go quiet. I ran the same infrastructure for my own operation first, so every send is measured end to end and yours arrives on the same instrumentation.
Own-domain send funnel: a 1×1 open pixel and tracked redirects join into GA4 by reference parameter, with a key-gated JSON report per campaign.
Read-time bot handling: a scanner classifier catches mail-gateway prefetchers and re-classifies at read time, so tightening one signature cleans the whole history with no migration.
Zero-dependency analytics: a GA4 Data API client with no SDK and self-minted auth, resolving live to cache to snapshot so it never errors, and computing a 5-step site funnel from Sessions to Engaged to Page end to CTA click to Call booked.
Observed-only verification: SMTP is checked directly, a provider confidence score does not count as an observation, and catch-all domains disqualify.
Fail-closed send gate: pre-send checks run against the rendered artifact, with duplicate suppression and leak scanning.
Mailbox-native telemetry: send, reply, and bounce ground truth is read from the mailbox itself, kept apart from the pixel data.
GA4analytics data source
I build the channels a company owns outright: organic search, a content library, and a house email list. Each one is an audience the company holds directly, and I build it to keep producing without a media budget attached.
Aeternity content system: 42 dev docs, 16 contributor guides, 9 landing pages, 89 blog and PR pieces, content duplication cut 60%.
I design how the product is packaged and priced: the tiers, the economic model under each one, and any points or credit system inside it. Pricing is something I build and ship, not a field someone fills in later.
I run the tests that decide what ships: A/B on copy, creative, and funnel steps, each read against a real control. The outcome of a test is a number, and that number is what moves to production.
[02] WHAT I DO
[03] HOW I MEASURE
Every action a user takes fires an event I log. Each figure I report is a count of those events, never an estimate, and I can open any one to see the event under it.
I wire this measurement into everything I run: campaigns, funnels, products. This site is one live instance of it, my own operation, wired the same.
EMAIL OPEN
In the report, an open is the weakest signal I keep, because a scanner can fake it.
[04] WHERE I START
I start any account by finding out what is true in it. I read the spend against the tracking, check which numbers have real events under them, and leave the budget alone until that pass is done.
On this account the ads ran with no measurement under them, and no install traced back to the ad that produced it. I audited the spend, then rebuilt the measurement until I could name what each dollar bought.
The report I inherited showed the money leaving the account and nothing about what it brought back.
[05] HOW AI RUNS INSIDE IT
AI runs the production work of my growth machine: the content, the localization, the ad creative, the sourcing, the analysis. A verification gate stands between every AI output and anything a customer ever sees.
This is the growth work AI runs for me in production today. The cost math, and the full story of running a whole company on AI, lives on my AI page. Here you get the lines that ship growth.
Campaign operations and content
I built Entangler and Helix, a company's first AI-native marketing infrastructure: campaign operations plus content generation, watching cross-chain data for anomalies.
EntangleSHIPPED · Campaign content, live on its own infrastructure.
Multilingual localization
An LLM localization pipeline that ships one company's content across 16 languages.
KuCoinSHIPPED · Localized content, live across 16 languages.
Ad creative from the live product
A pipeline that films a live app and cuts finished video ads, with personal data scrubbed out.
A consumer-fintech app, retail investors, iOSSHIPPED · Finished ad cuts, ready to run.
Creator and KOL sourcing
A crawl that finds the right creators by reading what they make.
A consumer-fintech app, retail investors, iOSSHIPPED · A vetted shortlist worth paying.
Lifecycle and outbound
LLM-personalized outbound, drafted per segment and gated before send.
My own operation; a consumer-fintech app, retail investors, iOSSHIPPED · Sent to verified recipients on my own stack, and designed as a full lifecycle architecture for a client.
Activation scoring
Inside a quest and activation loop, deterministic scoring rules rank real actions for authenticity and route cohorts, while code settles the rewards.
SpreeSHIPPED · Clean activation events and routed re-engagement cohorts.
[06] HOW GROWTH CHANGES BY BUSINESS
THE SAME MACHINE, BENT BY EACH SHAPE.
EXCHANGE AT SCALE
ACTIVATION AT POPULATION SCALE
PROBLEM · At exchange scale the funnel is arithmetic. Millions of users move through it, so a fraction of a point on first-deposit conversion is real money, and every region measures itself differently. The job is reading one funnel across languages and markets without losing the thread.
MACHINE · I instrument the whole funnel before anything else, unify the view across regions, localize the content at volume, and build the education and activation layer that turns a signup into a first deposit. Community retention runs off the same instrumentation.
RECEIPT · KuCoin: designed and launched KuCoin Learn end to end. Bitazza: the first unified view of funnel performance across regions.
[07] GROWTH RECORD
I put every figure from the cases above in one place, once each, so you can check them.
Education-led activation at exchange scale: KuCoin Learn, first-deposit programs, and the bot community, instrumented end to end.
FAQ
These are the doubts still worth raising after the numbers. I answer each one straight and point to where the page proves it.
The record already spans an exchange, a fintech, an NFT platform, and a protocol, and I carried the same growth machine across all of them. Every product has users to activate and money that has to work, so the funnel and the economics carry over whatever the category.
I build the systems: instrumented acquisition, activation loops, owned channels, and the pricing model underneath. Paid media has a place inside that, and I run it where the math works, but it rents attention that ends the moment the budget does. What I build stays with you and keeps producing after the work is done.
Activation is the first action that shows the product worked for a person, like a funded account, a verified action, or a completed setup. A signup only proves the account exists. I build the loop that carries people from the account to that first real action and instrument every step, so activation becomes a number you can watch and move.
I bring the measurement layer and already run it on my own operation, so your team skips the tracking project most teams start with. I wire the events, the attribution, and the reporting myself, then hand over a funnel view the team can read directly. In return I need access and honest answers about how the product works.
I build the systems and the instrumentation to be operated by the team that owns the product, working with whoever is already there. The owned assets, the playbook that drives them, and the numbers all live with you, readable by whoever runs them day to day.
The first move is an audit. I instrument what is already there, map the real funnel end to end, and find the stages where users are leaking before I build anything new. From that read I set the order, so the first thing I change is the thing that moves the most.