00

Ilia Dereviashkin

I build frontend products that turn complex workflows into reliable user experiences.

At Kameleoon, I help turn ambitious product requirements into customer-facing experiences - from expanding Feature Flags into a broader experimentation platform to taking an AI-assisted experiment builder from MVP to production.

Core
React / TypeScript
Company
Kameleoon · 2021—Present
Work
Remote
01

Kameleoon

AI-powered A/B Experiment Builder

Helped take an AI-assisted A/B experiment builder from MVP to production. The product became a major strategic direction for Kameleoon, was adopted by existing customers, helped attract new customers, and reduced the time required to create experiments.

Business impact & ownership

As the sole frontend engineer during the MVP, I owned the product-facing frontend delivery for prompt-based variation creation, human-in-the-loop questions, preview, compare, and manual editing - turning product requirements into one cohesive customer workflow.

Today, I continue to make most product-facing frontend design decisions in a three-person frontend team, collaborating with a frontend-oriented tech lead, backend engineers, and product management.

Customer workflow

  1. Describe the experiment
  2. Answer HITL questions
  3. Generate a variation
  4. Preview the result
  5. Compare alternatives
  6. Refine code manually
01

Own the end-to-end workflow

Designed the core interaction model so users can move from an idea to a reviewable variation without leaving the builder.

02

Keep product behavior predictable

Designed state and data boundaries across multiple services so conversation history, experiment data, generated code, and UI state stay consistent as the workflow evolves.

03

Make new capabilities easier to add

Introduced structured human-message content that later supported richer HITL interactions without redesigning the conversation-storage service.

04

Protect integrations as the product scales

Used generated API contracts and runtime validation to make service boundaries explicit and reduce integration risk as more product capabilities were added.

02

Kameleoon

Feature Flags & Experimentation

Joined when the product was primarily a feature kill switch and contributed as it expanded into a broader feature-management and experimentation platform, supporting more advanced rollout, targeting, measurement, and experimentation workflows.

Product expansion

Feature Flags
├─ Variations & variables
├─ Audience targeting
├─ Scheduling
├─ Goals & experimentation
├─ Results
└─ Integrations

My focus was keeping the customer experience stable and predictable while the product scope kept growing. The editor had to absorb new capabilities without becoming fragile or difficult to evolve.

As my responsibility grew, I became independent in frontend technical decisions and eventually operated as the sole frontend engineer for a period, covering feature delivery, maintenance, reported issues, and GraphQL API contract discussions.

  1. 01
    Started on the core product

    Joined as a junior while Feature Flags was still primarily a kill-switch product.

  2. 02
    Supported product expansion

    Delivered frontend work as targeting, scheduling, experimentation, results, and other customer workflows were added.

  3. 03
    Assumed frontend ownership

    For a period, independently covered the frontend side of the product and its ongoing delivery.

  4. 04
    Moved to the new strategic product

    After another frontend engineer joined Feature Flags, moved to AI Builder as its sole frontend engineer for the MVP.

03

Principles I use to keep complex frontend systems predictable.

01

Keep state minimal

Persist the source of truth. Store only the state required to reproduce the interface.

02

Derive instead of synchronize

Avoid duplicated representations of the same information. Derive them from authoritative state whenever possible.

03

Make boundaries explicit

Strong types inside the application. Generated contracts and runtime validation at system boundaries.

04

Refactor before complexity compounds

Small structural changes are cheaper while the system is still understandable.

05

Design for evolution, not speculation

Create room for demonstrated directions of change without generalizing for hypothetical requirements.

04

July 2021 — Present

Kameleoon

Middle Frontend Developer

Over 5+ years at Kameleoon, I progressed from junior contribution to independent frontend ownership across two customer-facing products. I supported the expansion of Feature Flags into a broader experimentation platform, then joined AI Builder as the sole frontend engineer for the MVP and helped take it to production.

2021
Feature Flags foundation

Joined while the product was primarily a feature kill switch.

Later
Expanded customer workflows

Delivered frontend work as the product added variations, targeting, scheduling, experimentation, results, and integrations.

Then
Sole frontend responsibility

Independently covered frontend technical decisions, delivery, maintenance, and reported issues for a period.

MVP
AI Builder launch

Joined the new product as sole frontend engineer during the MVP and helped take it from MVP to production.

Now
Strategic product growth

Continue to own most product-facing frontend design as the AI Builder team and product scope grow.

05

Tools are secondary to the system they serve.

Core

React · TypeScript · GraphQL · OpenAPI · Zod

Libraries & tooling

Redux Toolkit · Zustand · TanStack Query · React Router · Vite · Vitest · Playwright · Storybook

06

Remote frontend opportunities

Interested in working together?

I’m open to remote frontend engineering opportunities where product impact, ownership, and execution quality matter.

dereviashkinilia@gmail.com