Full-Stack Software Engineer

From idea to production.
Serious about quality.

I design and build web, mobile, and AI-powered products. My work spans architecture, implementation, testing, and production support, with AI integrated throughout the development process.

Explore my workExplore engineering workProduct, AI, and engineering-tooling case studies.

My projects

Explore my work.

Products built for real users and teams—from the first workflow to a reliable production handoff.

There is no web version of this app.

The other projects here embed a real Web build running against its production API. Symply House is built for iOS and Android only — it was never compiled to run in a browser, so there is nothing to put in this frame. Nothing is being held back; the app simply does not exist in a form a browser can open.

The engineering teardown beside this panel goes to the same depth as the others.

AI project guidePrepared, source-linked walkthroughs

Preparing the demo session…

Offline-First / Privacy/login

Household tasks, plans and notes stay useful when the connection drops. Local-first state lets people browse and update the home offline, then sync deliberately when the network returns.

ChallengeKeep a shared home usable without a connection

Wi-Fi is not guaranteed when someone is checking a task, opening a project or reviewing a floor plan. The app should keep the household workflow available offline while keeping private home data inside its account and device boundary.

DecisionMake the device the first place data lives

Read and write the working household state locally, queue changes when offline and make sync status explicit. For the portfolio, sign in with a shared public demo account — pre-filled by the app from a flag in its own URL, not by the portfolio — and keep the synthetic content in that browser tab alone, so no personal household records are exposed.

SolutionOffline workflows with controlled sync

The app stays navigable and actionable without a network, then reconciles account-scoped changes when connectivity returns. The backend supports sync and identity; it does not turn the public portfolio into a window onto private homes.

ImpactThe workflow survives the network

People can check plans and update household work offline, then reconcile changes when connected. The result is a calmer, more private product boundary: local work first, controlled sync second.

Implementation

local household store + queued mutations + visible sync state + origin-checked guest bridge

Expo RouterReact NativeLocal-first syncWorker auth

Ask the project guideInteractive, prepared answers grounded in the project evidence above.

Engineering case studies

Selected engineering work.

Products I helped build, decisions I owned, and what changed as a result.

PRODUCT ENGINEERING · STEP

I helped build Circles and the Movement Feed, connecting mobile experiences with an event-driven backend for memberships, invitations, posts, comments, and reactions.

My contribution: Product implementation across iOS and backend, with shared contracts and release planning.
Key decision: Keep access rules on the backend and make feed updates resilient to duplicate or missed events.

In this case studyOne correctness bug, traced from a missed event type to reconciliation and backfill

Read case study

APPLIED AI · STEP

AI coaching in production.

I helped build AI coaching features with contextual retrieval, streaming responses, and tool use. One key change was moving retrieval out of conversations that did not need it.

My contribution: AI integration, backend and client delivery, output validation, and performance investigation.
Key decision: Use retrieval as a tool instead of an unconditional step on every message.

Author-reported estimate, not a measurement~300–500 ms of retrieval overhead removed from about 90% of messages

Read case study

ENGINEERING TOOLING · STEP

Testing the agents that review engineering plans.

I built a structured review workflow and regression cases from real failures in AI-assisted planning. The goal was to check what the reviewers actually catch—not just whether they produce convincing feedback.

My contribution: Review protocol, reusable skills, evaluation cases, and verification rules.
Key decision: Turn recurring review failures into regression cases instead of adding untested prompt rules.

Recorded run · 2026-05-2811 regression cases: 6 pass, 2 fail, 2 partial, 1 fixture defect

Read case study

What colleagues say

People I’ve
worked with.

Recommendations from teammates and collaborators.

Kentaro KojimaFull-Stack Developer | Next.js, Go, AWSSeptember 12, 2026 · Worked with Andrei on the same team

He has a remarkably proactive, self-driven approach: he would quickly prototype AI-powered features, propose them to the team, and ship several of them into the product.

Read full recommendation

Irene ZhuMarketing & Business Development Leader | Global Brands, Consumer Growth & Partnerships | Digital, AI & Performance MarketingSeptember 11, 2026 · Worked with Andrei on the same team

He is incredibly detail-oriented, efficient, creative, and thoughtful in the way he approaches his work.

Read full recommendation

Andrei ChernykhSenior Software EngineerSeptember 10, 2026 · Was Andrei’s mentor

What I appreciate most is that he takes feedback seriously without taking it personally.

Read full recommendation

Olena TomanovaSoftware QA Lead. Scrum Master.September 10, 2026 · Worked with Andrei on the same team

He has a rare talent for turning complex requirements into clean, scalable, and reliable code.

Read full recommendation

Tandin WangchenFull-Stack Software Engineer | React, Node.js, TypeScript, AWS/Azure | Building scalable web apps & AI-integrated toolsSeptember 9, 2026 · Worked with Andrei but on different teams

He helped me a lot while I was learning Swift, especially with Xcode and AWS architecture, and explained things in a clear, practical way.

Read full recommendation

Let’s talk

Let’s build something reliable.

I’m interested in full-stack, applied AI, AI-enabled development and quality-focused engineering roles.

Email me

LinkedIn GitHub

From idea to production.Serious about quality.

Explore my work.

Selected engineering work.

Building the social layer at Step.

AI coaching in production.

Testing the agents that review engineering plans.

People I’veworked with.

Let’s build something reliable.

From idea to production.
Serious about quality.

People I’ve
worked with.