Skip to content
View Diizer's full-sized avatar

Block or report Diizer

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Diizer/README.md

Dmitry Zeleniak — Senior Full-Stack Engineer

Go · Node.js · Angular · React · AI coding agents

I build billing, payments and booking products end to end, and I build them with AI coding agents. I share what actually works: harness setup, agent orchestration and token economics.

Experience: 9+ years in software engineering
Domains: telecom billing SaaS, payments, booking platforms, third-party integrations
Open to: remote full-stack roles and freelance collaborations

LinkedIn · Upwork

What I do

  • Ship products end to end. Angular and React frontends, Go and Node.js services, SQL databases, CI/CD.
  • Modernize codebases that are already in production. Recent example: an eight-phase migration of a legacy React 16 portal to React 19, TypeScript, Redux Toolkit and Vite, including its first automated tests.
  • Integrate with everything. 20+ integrations across accounting and PSA (Xero, ConnectWise, Autotask, HaloPSA), tax (Avalara, CereTax) and payments (Stripe, Authorize.Net, Nuvei, GoCardless).

How I work with AI agents

I use Claude Code and Codex every day, and I treat the setup around them as engineering work of its own.

  • Context agents can actually use. Modular project guides loaded on demand, one source shared by Claude Code, Codex and Cursor, plus scripts that fail CI when the code drifts from the documented conventions.
  • Custom subagents and guard hooks. A subagent that keeps a live dashboard for long agent tasks, with token usage per step, and a hook that limits what that subagent can touch.
  • Fewer tokens for the same work. Code knowledge graphs, so agents query a graph before they grep and read whole files; effort and turn limits chosen by task type.
  • Agents in parallel. Separate branches in git worktrees, one agent per task.

I define the scope, review the output and verify behavior. Architecture, correctness and security stay my responsibility.

What I'm working on in public

  • claude-code-live-dashboard: a subagent that keeps a live dashboard for long agent tasks, and the hook that fences it in.
  • Measuring token usage with and without a code knowledge graph on a real codebase.
  • Learning herdr for orchestrating agent sessions, and writing down what works and what does not.

Stack

Area Technologies
Frontend Angular, React, Next.js, TypeScript, RxJS, Redux Toolkit, AG Grid, Tailwind CSS
Backend Go, Node.js (Express, NestJS), Python (Django REST Framework), REST, OpenAPI
Data PostgreSQL, Microsoft SQL Server, Redis
Delivery AWS, Docker, GitHub Actions, Bitbucket Pipelines
Testing Playwright, Vitest, Mocha, Cypress
AI engineering Claude Code, Codex, subagents, hooks, skills, context engineering, token optimization

Selected repositories

  • claude-code-live-dashboard — Claude Code subagent with a live progress dashboard, token usage per step and a guard hook.
  • voice_poc — Python speaker-verification prototype with voice activity detection.
  • languages-cards — language learning flashcards built with Vue 3, TypeScript and Pinia.
  • v0-rm-autoturbo-website — multilingual Next.js site in four languages.

Have a product to build or a codebase to improve? Connect on LinkedIn or reach out through Upwork.

GitHub stats

Pinned Loading

  1. languages-cards languages-cards Public

    A modern language learning flashcard application built with Vue 3, TypeScript, and Pinia.

    Vue

  2. Rasterex-Web-Viewer Rasterex-Web-Viewer Public

    Forked from ByteHandyman/Rasterex-Web-Viewer

    Angular Webviewer Source Code

    JavaScript

  3. v0-rm-autoturbo-website v0-rm-autoturbo-website Public

    Multilingual Next.js / TypeScript website for RM Autoturbo, built with v0. English, Catalan, Spanish, and French.

    TypeScript

  4. voice_poc voice_poc Public

    Python speaker-verification prototype: voice activity detection, short utterances, and background speech handling.

    Python

  5. claude-code-live-dashboard claude-code-live-dashboard Public

    A Claude Code subagent that keeps a live progress dashboard for long tasks, plus the hook that fences it in.

    JavaScript