Diagnose first
I do not treat symptoms. I break a metric into a tree and look for the branch that actually moves the result.
At MTS Bank I built the product metrics tree that feature launches were judged by and the backlog was prioritised with.
Open to offers: remote or relocation
I turn noise into products.
Four years in fintech: T-Bank, MTS Bank, lead generation for the US market. I find the signal in the data, build a product around it and take it to production. This past year I have done it with AI agents too: five products of my own, from PRD to code.
Product Manager · fintech and AI productsMoscow · working remotely

Before and after. Warm is before, cool is after.
before: 31%after: 75%
application-to-sale conversion, FINLEADGEN+142%
before: 9%after: 30%
digital share of sales, MTS Bank, 2024+21 pp
before: 35%after: 50%
landing view to application+43%
before: 1.7after: 2.5
products per corporate client×1.5
5months
from concept to a production platform
5
products of my own, built with AI agents

I do not treat symptoms. I break a metric into a tree and look for the branch that actually moves the result.
At MTS Bank I built the product metrics tree that feature launches were judged by and the backlog was prioritised with.
I build the analytics first and argue second. A hypothesis without a measurement is an opinion with good slides.
At FINLEADGEN I built the analytics on PostgreSQL and Metabase myself; lead-routing decisions rest on it.
I record every notable decision as an ADR: context, options, the choice and what it costs.
50 ADRs across five products of my own, including the ones I later had to reverse.
Agents write the code; I own the product: the spec, the trade-offs and the quality gates a release has to pass.
Rezana: 1,289 automated tests across the backend, the frontend and the API contract.
Two fintech jobs and five products of my own. Hover a row and the field takes the product’s shape.
Every job is a version. A major one when the role or the market changed.
Agents write the code. I own what would not exist without me: why, for whom, what counts as success and where the quality bar is.
Market, personas, competitors and unit economics, before the first line of code.
What we build and what we don’t. Success criteria in numbers.
Every notable choice is an ADR: options, cost, triggers for revisiting.
Claude Code under the project’s rules: CLAUDE.md, skills, hooks and documentation that updates itself.
Tests, types, linters and product gates like presence ≥ 0.5 in Sample Zero.
Every push to main builds and ships to testers. Status is written down honestly.
Counted by a script across the repos on 23 September 2026
Plekhanov Russian University of Economics, Moscow
B.Sc., Applied Informatics in Economics, 2026
Charles University, Pilsen
Medicine, 2020
Did not finish. It left me the habit of diagnosing before treating.
It knows what this site says, and nothing more. If it does not know, it will say so.
Looking for a Product Manager role in fintech or AI products. Remote or with relocation.
In Moscow it is ——:——I usually reply within a day