Selected work

Projects

Case studies focused on context, ownership, architecture, technical decisions, and what each project demonstrates about senior-level engineering work.

Current / Core platform

AM Commerce Platform

Reusable commerce platform foundation

Python

FastAPI

SQLAlchemy 2.0 async

PostgreSQL

Alembic

Redis

OAuth2 / JWT

RBAC

Domain events

Outbox pattern

DDD

CQRS

Docker

GCP Cloud Run

OpenTofu

Architecture proof

AM Commerce module map

A concise public map of the core platform architecture: FastAPI async backend, PostgreSQL, SQLAlchemy 2.0 async, Alembic migrations, RBAC, Outbox, notifications, Unit of Work, and DDD/CQRS direction.

Core backend

FastAPI asyncPostgreSQLSQLAlchemy 2.0Alembic

Patterns

RBACOutbox patternUnit of WorkDDD / CQRS direction

Shared libs

libs/fastapi-corelibs/domain-utilslibs/rbaclibs/outboxlibs/notifications

Jobs / runtime

db-migrations-joboutbox-processor-jobDockerGCP Cloud Run / Cloud SQL directionOpenTofu direction

Modules

auth-gateway-apiusers/admin services plannednotificationsrepository excerpts available privately

Public architecture mapPrivate repository evidence available on requestCode excerpts can be shared when appropriate

Read the platform-first architecture note

01 / Context / Problem

AM Commerce Platform is the reusable commerce foundation behind marketplace products. It is the platform layer for backend services, shared modules, permissions, events, notifications, integrations, and operational workflows.

02 / My role

Founder, lead engineer, and system architect. I own the backend foundation, shared platform boundaries, service model, and infrastructure direction.

03 / What I built / architecture

FastAPI backend with routers, services/use cases, repositories, domain models, auth, RBAC, domain events, outbox processing, notifications, and infrastructure adapters. PostgreSQL, SQLAlchemy 2.0 async, Alembic, Redis, and clear API contracts keep the platform reusable.

04 / Technical decisions

Platform-first design is slower at the beginning, but it keeps marketplace products from becoming one-off implementations and makes shared primitives reusable.

05 / What this demonstrates

Authentication, RBAC, user and permission management, internal APIs, shared modules, outbox processing, notifications, integration boundaries, and production-oriented infrastructure are treated as core platform concerns.

06 / Current status

Current core platform work. No fake scale, revenue, traffic, or user-count claims.

MVP in active development

Armenian Mall

A founder-led marketplace MVP for the Armenian market, powered by AM Commerce Platform and designed around users, sellers, products, permissions, notifications, and future operational workflows.

AM Commerce Platform

React

Next.js

FastAPI

PostgreSQL

Tailwind CSS

marketplace modules

admin tools

Product proof

Marketplace MVP flow

A neutral product-flow map for Armenian Mall as a marketplace MVP built on top of AM Commerce Platform. No user, sales, traffic, or GMV claims are made.

Seller/admin flow

Seller/AdminProduct managementCatalogOrdersNotificationsAM Commerce Platform

MVP roadmap

Auth & rolesCatalog foundationSeller/admin flowsOrders / checkout directionIntegrations later

Evidence available

Product-flow mapSeller/admin workflowMVP scope record

Product-flow mapMVP roadmapDemo available when public scope allows

Read the MVP execution note

01 / Context / Problem

Armenian Mall is a founder-led marketplace product for the Armenian market. It is the product and business use case built on top of AM Commerce Platform.

02 / My role

Founder, product owner, and lead full-stack engineer responsible for marketplace workflows, product direction, and implementation on top of the platform foundation.

03 / What I built / architecture

Armenian Mall uses AM Commerce Platform for users, sellers, products, permissions, notifications, integrations, and future operational workflows while keeping marketplace-specific behavior in the product layer.

04 / Technical decisions

The product stays connected to AM Commerce Platform instead of becoming a separate unrelated system, while product-specific seller, catalog, admin, and operational needs remain explicit.

05 / What this demonstrates

The work focuses on a credible marketplace MVP for the Armenian market: seller/product workflows, admin needs, permissions, notifications, and future operations without pretending the product already has public scale.

06 / Current status

MVP in active development. No fake user, revenue, traffic, or marketplace scale claims.

Completed / Confidential

Confidential Commerce Integrations & Automation

Integration-heavy commerce automation involving marketplace APIs, 1C workflows, Telegram-based operational tools, scheduled data synchronization, and internal business automation.

FastAPI

React

PostgreSQL

Ozon API

Wildberries API

Yandex Market API

CRM integrations

Payment APIs

Telegram bots

ETL pipelines

Scheduled synchronization

Sanitized workflow

Sanitized integration workflow

A confidentiality-safe workflow diagram for marketplace/API automation work. Client identity and private business details stay hidden.

External systems

Ozon APIWildberries APIYandex Market API1C integrationTelegram automation

Internal workflow

Scheduled/background jobsData synchronizationNormalization layerRetriesIdempotency

Operator visibility

Monitoring/admin visibilityNotificationsTelegram bot actionsSanitized code excerpts when allowed

Sanitized workflow diagramRetry/idempotency mapPrivate implementation details protected

Read the marketplace integrations note

01 / Context / Problem

A confidential commerce client needed to connect marketplace operations, internal workflows, and business automation without exposing business details publicly.

02 / My role

Backend, full-stack, and integration engineer working across marketplace APIs, operational automation, data synchronization, and internal tooling.

03 / What I built / architecture

Integrations with Ozon, Wildberries, Yandex Market, 1C, CRM systems, payment providers, Telegram bots, ETL pipelines, scheduled jobs, imports/exports, reporting, supplies, adverts, orders, and stock analytics.

04 / Technical decisions

The work favors maintainable synchronization and operational reliability over exposing business details or inventing public-facing metrics.

05 / What this demonstrates

Confidentiality, external API behavior, data consistency, partner failures, scheduled synchronization, and Telegram-based operational workflows shaped the implementation.

06 / Current status

Confidential case study. Client name and business details remain private.

Planned / Founder-led educational platform

Coding Armenia

A founder-led educational platform and developer community focused on practical programming courses, tutorials, multilingual-ready content, SEO-aware routing, and course-track architecture.

Python

FastAPI

PostgreSQL

React

Next.js

Tailwind CSS

SEO

Docker

GCP

Content model proof

Course-track content model

A lightweight education-platform proof map for course structure, content pages, blog/articles, and founder-led product direction.

Course tracks

BackendFrontendFull-StackCybersecurity & NetworkingData Science & AIDevOps & Cloud

Content model

CoursesLessonsBlog / Engineering NotesProjectsAdmin/content management

Roadmap

Course tracksContent pagesBlogAdmin/content managementAuthentication later

Course-track diagramContent model mapRepository/demo scope pending

Read the product engineering note

01 / Context / Problem

Armenian-speaking developers need practical programming education, course tracks, tutorials, and technical resources that are structured as a real product rather than scattered content.

02 / My role

Founder and lead full-stack engineer responsible for product structure, backend foundation, frontend experience, roadmap planning, and educational content direction.

03 / What I built / architecture

Planned platform with Python, FastAPI, SQLAlchemy 2.0, PostgreSQL, React, Next.js, Tailwind CSS, Docker, GitLab CI/CD, GCP, OAuth2/JWT, SEO-optimized routing, multilingual structure, and course-track architecture.

04 / Technical decisions

The first useful version should stay focused on content structure, SEO, and clear learning paths before adding heavy community or account features.

05 / What this demonstrates

Backend, frontend, full-stack, DevOps, AI, and cybersecurity tracks need a coherent content model, brand system, design system, and roadmap without pretending the product is already mature.

06 / Current status

Founder-led educational platform and developer community in early concept/platform planning.