- π οΈ Software engineer with 3+ years shipping web apps in SaaS, EdTech, FinTech, Web3.
- π At MetaLamp: lead frontend on a banking web app, an investment mobile app, and a TON-based DeFi Mini App for Telegram.
- π At Flearn (EdTech): led a small team and shipped an LMS MVP in 3 months.
- π― Help teams cut time-to-market by 30β50% through pragmatic delivery and modern tooling.
- π Open to product teams working in Web3, FinTech, EdTech with React/TypeScript and AI-augmented workflows.
I don't just use AI tools β I build the workflow around them. My current dev loop is a multi-agent pipeline I designed and iterate on continuously.
- Tools in daily use: Claude Code, Codex CLI, local LLMs via Ollama (qwen for QA), Playwright for browser automation.
- Custom workflow (
/impskill): plan β skeptical Codex review β autonomous implementation β independent audit β hypothesis-driven QA β PR β fully orchestrated, with Telegram approvals at risk gates. - Built-in safety: mandatory human approve for
mainmerges, fail-closed risk gates, recall/recap memory layer so context survives across sessions. - Self-improvement loop: the agent runs scientific cycles (observe β metric β hypothesis β experiment β verify) on its own delivery, scheduled autonomously β measurable improvements per cycle.
- Scope: the goal is fully remote operation β voice + Telegram, zero typing β with the agent reliably delivering and reporting.
Day-to-day + production work across MetaLamp / Flearn projects (most of it lives in private/company repos β see Featured public work below for what's open).
Languages
Frontend
Web3 & FinTech
AI / Agents / Tooling
Most production work lives in private/company repos. Public side projects:
- π οΈ lexer β TypeScript Β· build-your-own programming language: tokenizer + parser experiments.
- π¬ messenger β TypeScript Β· realtime messenger sandbox.
- π Thesis β Python / Jupyter Β· forecasting weather derivatives with LSTM RNNs (master's thesis).
- π° Finance β Python Β· quick-look on company financial performance.
A collection of standalone open-source skills for Claude Code β covering the full AI-assisted dev cycle:
| Skill | Description |
|---|---|
| claude-skill-imp | Full dev-cycle orchestrator: plan β review β iterate β audit β PR |
| claude-skill-telegram | Telegram notifications and inbox integration |
| claude-skill-slack | Slack diagnostics and draft integration |
| claude-skill-tracker | Yandex Tracker task management |
| claude-skill-qa-agent | Hypothesis-driven QA with Playwright |
| claude-skill-planner | Big-task implementation planner |
| claude-skill-review | Code review (PR/branch/commits) |
| claude-skill-fix-audit | Lightweight audit for quick fixes |
| claude-skill-iteration-implementer | Iteration implementer from plan.md |
| claude-skill-auditor | Audit iteration results against ACs |
| claude-skill-session | Local model (Ollama) reference guide |
Install: git clone https://github.com/VladimirFyodorov/claude-skill-<name> && bash claude-skill-<name>/install.sh
β‘ Open to interesting product teams. The fastest way to reach me β Telegram.



