Vova Zakharov

Vova Zakharov

Helping our future overlords walk since 2020

Fractional CTO for teams that don’t want to YOLO into the agent era.

A lot of teams now have agents writing their code, and nobody whose job it is to keep what comes out shippable. That’s the job I do: I draw the architecture, set up the pipeline the agents ship through on your codebase, and sit in the pull requests rather than in the org chart. Last time roundLast time round that took a live product from a no-code builder to 250,000 lines of TypeScript, tests included, in 158 days — and the engineers who took it over run it today.

  • Idea → production
  • Architecture and foundations
  • Your team on agent rails
  • Rescue and diagnosis
  • Standing technical judgment
Read full CV

Featured Projects

Playgram.ai

Read the case study →Read the case study →

A live, feature-rich AI chat product lifted off a no-code builder into a production Next.js codebase in 158 days, and in production for its users the whole way through.

Next.js 16, Supabase + Drizzle, Weaviate, LiteLLM, Railway, feature-sliced design

muthur

The engineering platform that rebuild ran on, extracted so it travels: the staged agent pipeline and the pre-push gate, carried onto any project — and kept in sync, so a fix made in one repo reaches the others.

Claude Code, agentic development, open source

jukebox-webui

★ 84

Suno before Suno: running OpenAI's music model in Google Colab back when there was no product to use instead.

Python, Gradio, Google Colab, open source

almostmagic

★ 65

Structured generation before it was a feature — typed output from a single call, years before every SDK shipped its own version of it.

TypeScript, OpenAI API, npm package, open source

Same pattern, earlier: write (a BYOK AI-first text processor), mindy (ChatGPT before ChatGPT), ideality-nuxt (no-code AI widgets).

Work Highlights

Playgram (March–August 2026)

Rebuilt a live AI chat product from Bubble into production Next.js 16 in 158 days: 250,000 lines of TypeScript with tests included, none of it hand-written, up to 20 agents working at once, a deploy every 2.4 days. Reviewed and mentored three engineers on the platform, which they run today without me.

Next.js 16, Supabase + Drizzle, Weaviate, LiteLLM, Railway, feature-sliced design

DDB / randddb.com (2023-2025)

Experimental AI platform at one of the world’s largest ad firms. Built enterprise-scale suite:

  • Chatbot: ChatGPT-style corporate assistant with documents, images, personas, multi-user support
  • Robowriter: Build-your-own-agent content generation tool
  • Picmaker: Replicate-powered image generation/training
  • YesYouCannes: Cannes award deck generator (idea → imagery in minutes)
  • Spaces: White-label mini-platforms for agency clients

Django + PostgreSQL, Vue + TypeScript

Orcool (June-August 2025)

AI-agent-based review intelligence tool for brand marketing. Built end-to-end: automated review collection → AI summarization (SWOT, competitive positioning) → SaaS platform.

Next.js/NestJS, Cloudflare Workers, Firebase, custom LLM orchestration framework

Read full CV

Contact

Email: vzakharov@gmail.com

GitHub: github.com/vzakharov

LinkedIn: linkedin.com/in/vovahimself

X/Twitter: x.com/vovahimself

Substack: substack.com/@vovahimself