Elena Tonkikh

I help product teams reduce uncertainty before development:
I turn complex B2B workflows into clear product logic, prototypes, requirements and measurable discovery outcomes

Focus — Product discovery · B2B SaaS · UX/CX background · AI-assisted prototyping

UX roots, eight years in product. Strongest in B2B SaaS, fintech and complex CX: I find where users lose context, frame the problem in business terms and carry it through to a product solution that reduces support load and cost-to-serve

Experience — 8 years
2026

Atomic English · solo pet project + AI agents

Pet project · Android MVP · research project

Tested whether one person with AI agents can ship an Android MVP end-to-end: product logic, UX, runtime, dictionary contract, QA and a managed build pipeline

2024 — 2026

Senior Product Manager · Product Owner

X-Keeper · B2B telematics · 2M+ devices

Shifted the customer dashboard from a support-heavy model to genuine self-service: rebuilt device-setup flows, introduced an asset-centric monitoring model and shaped the web and mobile experience for large B2B accounts

2020 — 2024

UX/UI → CX → Product Owner

Sber · fintech / loyalty · team of 15

Operated inside a large fintech environment: interfaces, design system, user research, CJM, hypotheses, backlog and adjacent teams

2018 — 2020

UX/UI designer · branding and web

Studios and freelance

Branding, landing pages, web interfaces and no-code prototypes. Systems thinking grew out of components, states and patterns

Education
2022 — 2023

HSE University · Analytics in Digital Product Management

2015 — 2020

Russian State University of Justice · Civil Law

Master's
Courses

Yandex Practicum · Product Manager · 2023  ·  Moscow Digital Academy · UX Design  ·  Paper Planes · Win The Market & Win The Digital  ·  UsabilityLab · Customer Journey Map training  ·  Krasnodar Art College · Easel painting

Inside this document
  1. 02 How I work · Agile role & Discovery Role · Discovery · AI
  2. 03 Case: From devices to assets X-Keeper · 2025
  3. 04 Case: Self-service instead of a support call X-Keeper · −20% support
  4. 05 Case: Atomic English · solo + AI agents Pet project · 2026
  5. 06 Case: Fintech, CX, 15-person team Sber · 2020 — 2024
Elena Tonkikh · How I work Role · Agile · Discovery · tonkikh.netlify.app →

My role in an Agile team — and the value
I bring at discovery

My core responsibility is discovery and product logic. In the later stages I step in selectively — I help preserve the intent of the solution, but I don't replace design, systems analysis, delivery, engineering or QA

01

Discovery

My role — core
  • Frame the business problem, user flow and success criterion
  • Build a flow prototype: steps, states, permissions, errors, edge data
  • Run the research, tests and leadership presentations
  • Stress-test the feature for viability before it reaches a sprint
02

Backlog refinement

My role — supporting
  • Write feature documentation
  • Split must-have scope, risks and future improvements
  • Help the team keep user logic intact while decomposing stories
03

Sprint delivery

My role — selective
  • Answer product questions whenever an implementation fork appears
  • Help the team trade off value, time and scope
  • Check the demo against the original flow — without managing engineering
04

Release & learnings

My role — product learnings
  • Check whether user behaviour and support load have shifted
  • Isolate the real effect from noise
  • Frame the next round of hypotheses and improvements
Discovery — that cuts the cost of being wrong

My job is to help the team see — earlier — which hypothesis works, where the flow breaks and what should not be handed to engineering at all. The team saves time, design and engineering stay unburdened, and discovery leaves behind a clear base of decisions

01 · Hypotheses

Validated before expensive engineering

I validate hypotheses on flow prototypes, real data and explicit success criteria

02 · Risks

Less wasted work

I pin down roles, permissions, states, errors, data and edge cases up front.

03 · Decision memory

Artefacts the team can return to

I leave behind reports, documentation, research findings and metrics for every hypothesis

How I use AI agents in discovery

AI-assisted prototypes that leadership and research can act on

Elena Tonkikh · Case 01 X-Keeper · 2025 · objects.html →

From devices to assets: a new remote-monitoring model for bank lessors

When new connections slowed, the business needed to grow differently — a second device on already-connected assets. But the product showed devices separately, and the second device's value wasn't visible. I rebuilt the product model: an asset is a bundle of devices, not their sum

+12%Assets with two devices
2MDevices in install base
1Unit of monitoring — the asset
01Context

Growth had to come from the second device

A crisis and rising key rate slowed financing of new vehicles. With fewer new assets, the business had to grow on already-connected ones — a second device per asset. The dashboard is the only remote-control surface once the device is installed.

02Problem

A model limit, not a UI gap

In the dashboard, devices appeared as separate entities even when they belonged to the same asset. Operators stitched events together by hand. Sales couldn't argue the value of a pair — because the model didn't show the pair

03Approach

An asset as one observable unit

Reframed the unit of monitoring from a device to an asset. Routes, events and states roll up to the asset; a second device adds confidence rather than noise. Sales got a coherent story; security got a bird's-eye view of the fleet

To show the value of a second device, the unit of observation had
to change first. — Reframing the task
B2B SaaSTelematicsAsset modelBanks & leasingOperatorsSales enablementWeb + Mobile2M devices
Elena Tonkikh · Case 02 X-Keeper · 2025 — 2026 · auto.html →

Self-service instead of a support call

On a B2B telematics platform, users were calling support for actions already available in the dashboard. I rebuilt the key flows so clients could configure devices, manage assets and pull data themselves — without a manager

−20%Support tickets on device reconfiguration
+9%Sessions in the customer dashboard
×4Mobile-app sessions
01Context

The dashboard wasn't a growth channel

Session Length wasn't a target metric. A good B2B dashboard should help clients finish tasks faster, not keep them inside the screen. Users could already configure devices and view events — but the flows were scattered and depended on hidden settings.

02Strategy

Independent work, end to end

Turn key flows into clear self-service inside the dashboard: visible next step, explicit dependencies between settings, and the ability to complete device setup, asset work or data retrieval without contacting a manager.

03Approach

Four product moves

  • Tabular view for large fleets — fast sorting beyond the map
  • No dead-end flows — system enables dependent settings
  • Timeline + second metric on the route — analysis, not viewing
  • Mobile relaunched on Kotlin Multiplatform
Fewer support tickets start with clearer product logic — not «fix the screen», but finish the scenario — Self-service principle
Self-serviceB2B SaaSDiscoveryCJMTabular UXKotlin MPMobile relaunchSupport load
Elena Tonkikh · Case 03 Solo Pet Project · 2026 · app.html →

Atomic English: shipping an MVP solo, with AI as a synthetic team

A full-stack product cycle led by one PM with AI-assisted execution. Narrowed the real-content idea into an Android MVP, built product logic, a Supabase/Postgres backend, a curated dictionary and a testable learning loop. The main takeaway: a learning product stands on trust in the data

14k+Meanings in the dictionary
300hProduct, dev and data cleanup
Solo + AIOne PM as orchestrator
01Insight

Vocabulary growth above A1 needs trustworthy data, not gamification

Built for an adult learner: basic English is already there. The next stage needs quality words, precise meanings, short context and a clear learning queue — not mascots and streaks.

02Pivot

From media discovery to a curated dictionary

Discovery with texts and audio surfaced words of uneven quality and added storage / sync / release risk. The MVP narrowed to a curated B2-level deck — proving the loop first, scaling sources later

01 · Find Words

Curated B2-level deck

Pre-curated quality words: diverse parts of speech, clear frequency, learning value.

02 · Word Card

Word, meaning, context

Translation, learner-facing meaning, and context tied to the selected meaning

03 · Lemmas

Learning queue

Selected words enter the queue and stay until a deliberate user action

04 · Chunks

A word in short context

Focus shifts from an isolated word to natural use.

05 · Review

Spaced review

The word returns on the right day and stays part of the system

Modern AI tools can ship MVPs and complex products — but there is no «make it good» button. What matters is management discipline and Agile principles: framing, ownership, context, validation, orchestration. — Outcome · Hypothesis confirmed
QR code to Atomic English beta-test instructions
Try the Android MVP
Scan to join the beta test
Elena Tonkikh · Case 04 Sber · 2020 — 2024 · sber.html →

Fintech product: interfaces, CX,
a 15-person team

Four years inside Russia's largest bank and fintech. I moved step by step from screens to product logic — investments interface design, then customer-experience work and hypotheses, then Product Owner in the Sber app team. The shift from thinking in screens to thinking in product systems

01 · UX/UI DesignerInvestments

Investment interfaces and design system

  • Designed components and screens for investment products: charts, showcases, product cards and learning materials
  • Built and maintained design-system components in Figma — web, iOS, Android, states and dark theme
  • Ran design reviews and refined work based on research and analytics
02 · CX ManagerCustomer experience

Customer journeys and hypotheses

  • Worked through customer journeys and located friction points
  • Ran target-audience, user-need and competitive analysis; generated and validated hypotheses
  • Built concepts, prototypes, CJMs and stakeholder materials; defended a strategy for a new cross-platform feature
03 · Product OwnerSber app · accounts & cards

Sber app: accounts and cards

  • Led accounts-and-cards flows with a 15-person team, adjacent teams and contractors
  • Owned the backlog, priorities and requirements; aligned product, design, analytics and development via Jira / Confluence
  • Built a unified onboarding to «Spasibo» for non-bank customers across partner surfaces
This period moved me from thinking in interfaces to thinking in product systems — the foundation for the later work on X-Keeper and Atomic English — Main shift in role