# Mikhail Shytsko

> Software & AI Engineer based in Bratislava, Slovakia. Solutions & AI Engineer at Slovenská sporiteľňa (Slovakia's largest retail bank, 2M+ customers) since July 2025. Founder of Seedfast — a schema-aware synthetic test-data generator for PostgreSQL for compliance-sensitive industries — since October 2025. 1st place at DDAccelerator Finals 2026 (Intelligent Digital Technology category, only Slovak startup among 9 countries).

## About

- [Portfolio](https://mikhail.shytsko.com): Single-page CV with summary, experience, projects, education, and contact links. Updated October 2026.

## Projects

- [Seedfast](https://seedfa.st): Schema-aware synthetic test-data generator for PostgreSQL. Reads a database schema and produces referentially consistent, realistic data in under 3 minutes — built for regulated industries (finance, pharma, medtech) where production data cannot leave secure environments. Generates realistic, referentially correct data in any locale and any domain; evaluation frameworks and automated correctness checks more than doubled output quality on held-out data. Ships a Go CLI and an MCP server for engineers and AI agents in production and CI, plus a web app with free Postgres tooling. 1st place at DDAccelerator Finals 2026.
- [DDAccelerator article on Seedfast](https://www.ddaccelerator.com/news/seedfast-makes-database-seeding-fast-clean-and-reliable-straight-from-the-cli)

## Experience

- Slovenská sporiteľňa, Solutions & AI Engineer (Jul 2025 — present): AI agent for credit review that cross-references lending, collateral and limits, replacing a manual multi-system lookup — approved for production as a Koog agent sandboxed in its own virtual filesystem, reaching bank data only through a least-privilege, auditable MCP tool layer with strict data minimization and a controlled disable path; change-detection that surfaces and explains material changes across products, limits, collateral and client data; LLM pipelines for automated credit-data validation and risk scoring on Azure OpenAI and Anthropic models, traced and evaluated with LangSmith, LangFuse and Ragas; 85% faster responses for two high-load services serving 2M+ customers; React/TypeScript features for a micro-frontend banking app; built a 3-person engineering team from scratch, doubling feature delivery.
- Seedfast, Founder & Founding Engineer (Oct 2025 — present): owns the generation core, Go CLI and MCP server, web app onboarding and lifecycle emails, containerised infrastructure and CI, plus demos and hands-on customer onboarding. See Projects above.

## Profiles

- [GitHub](https://github.com/xfiive)
- [LinkedIn](https://www.linkedin.com/in/mikhail-shytsko/)
- [X / Twitter](https://x.com/mikhailshytsko)
- [Hackernoon](https://hackernoon.com/u/mikhail-shytsko)
- [Linktree](https://linktr.ee/mikhsh)

## Expertise

- AI / LLM engineering: agentic LLM applications, MCP-based agents and tooling, RAG, LLM evaluation frameworks (LangSmith, LangFuse, Ragas), LangGraph, Koog
- Backend architecture: Spring Framework (Kotlin/Java), Python, Go, Apache Kafka, high-load microservices
- Frontend: TypeScript, React, Next.js
- Databases: PostgreSQL, schema-aware synthetic data generation
- Domain: banking and regulated-industry software

## Languages

- English (C1, IELTS Academic 8.0), Slovak, Czech, Russian, Belarusian