RU
Konstantin Larionov — fullstack developer and DevOps engineer in Voronezh

Over 4+ years I went from content manager and sysadmin to a developer who writes code, configures servers, and ships to production. I understand how a business actually runs — not just code, but the processes around it.

The past 1.4 years — deliberate practice with AI agents: context engineering, RAG systems, multi-agent orchestration, LLM integration into real business processes. I take a base model (Claude, GPT) and build the context around it so the model actually solves the client's task — not just runs a demo.

In parallel — fullstack on the Veloport team: backend, frontend, content pipeline with GPT copywriting across 22,000+ product cards, catalog support. My personal SaaS — Thumós , on Next.js.

Self-employed (Russia NPD tax regime), TIN 366606549484. Contract with both legal entities and individuals; a fiscal receipt for every payment via the "My Tax" app. Active status verifiable at  npd.nalog.ru .

4+years in IT
5projects in production
23k+products processed

Work history

Real backend tenure: enterprise 1С servers, PostgreSQL/MS SQL clusters, IS analytics. Not "took a course" — actually shipped.

  1. February 2025 — present

    Fullstack developer · AI integration

    Freelance · Larionov's Web Workshop

    • Fullstack on the Veloport team: Bitrix-based bicycle e-commerce, 100+ brands, 40+ catalog categories, content pipeline of 22,000+ SKUs with GPT copywriting.
    • Client web work: auto-service landing Niva36, local SEO landings, pro bono web showcase for scientific freeware.
    • AI integration: RAG systems, multi-agent orchestration, LLM integration into business processes (Anthropic SDK, Claude Code, OpenAI GPT API, MCP). 1.4 years of deliberate practice.
    • Personal SaaS Thumós on Next.js + Prisma + PostgreSQL: 18 validated psychological tests, forum, user accounts.
    • Automation: parsing (Python / BS4 / Playwright), 1С-Bitrix REST integrations, Excel catalogs of 23,000+ items.
    • DevOps: Linux servers, Nginx, Docker, PM2, SSH/rsync deploys, SSL.
    • Building a personal context-management framework for AI agents: long-term memory, sub-agent orchestration, autonomous task loops — applied in personal and client projects.
    Anthropic SDK Claude Code RAG MCP Next.js TypeScript Python 1С-Bitrix PostgreSQL
  2. October 2023 — February 2025

    Senior 1С support engineer (tier-2)

    MTS Digital

    • Resolution of complex IT incidents on enterprise infrastructure.
    • Administration of 1С server clusters and databases (PostgreSQL, MS SQL Server).
    • 1С code debugging, performance issue resolution.
    • Integration buses, PowerShell automation scripts.
    1С 8.3 PostgreSQL MS SQL PowerShell clustering
  3. December 2021 — October 2023

    Application systems administrator

    MTS Retail

    • User administration in Active Directory, Oracle, 1С.
    • SQL data analysis queries, integration bus monitoring (XML).
    • Routine task automation via PowerShell.
    Active Directory Oracle SQL PowerShell

References

Contacts available on request: TRIAR Group · MTS Retail · Veloport.

Stack by category

Not "familiar with" — what I actually use across the projects in portfolio .

AI / Context

  • Anthropic SDK
  • Claude Code
  • OpenAI GPT API
  • RAG
  • multi-agent
  • memory systems
  • MCP
  • prompt engineering

Frontend

  • React
  • Next.js
  • TypeScript
  • Vue.js
  • HTML5
  • CSS3
  • responsive

Backend

  • Node.js
  • Python
  • PHP (Bitrix)
  • REST API
  • WebSocket

Databases

  • PostgreSQL
  • MySQL
  • SQLite
  • MS SQL Server
  • Prisma ORM
  • SQL

DevOps

  • Linux (CentOS, Ubuntu)
  • Nginx
  • Docker
  • PM2
  • SSH/rsync
  • SSL
  • PowerShell

Integrations

  • 1С-Bitrix
  • 1С:UT
  • Bitrix24 REST
  • Telegram Bot API
  • YooKassa
  • parsing (BS4, Playwright)

SEO & content

  • schema.org
  • Open Graph
  • Yandex.Metrica
  • Google Analytics
  • product copywriting
  • SEO architecture

Got a project? Tell me — I'll reply within a day.

No phone calls, no 10-page proposals. Just text and facts.