I'm Roman, Technical Founder & AI Product Engineer.
After 10+ years in product & growth, I went all-in on building AI-powered products. I understand business logic and build like an engineer.
|
AI-first travel planner — itinerary generation, POI scoring, real-time recommendations, and an AI chat assistant that knows everything about your trip.
|
|
Tech blog about AI engineering — deep dives into LLM orchestration, edge functions, context engineering, and building production AI systems. Written from the trenches.
|
|
Use Gemini CLI from Claude Code to review code and delegate tasks. |
Interactive visual map of your Claude Code ecosystem: skills, agents, MCP servers — auto-discovered from .claude/ config. |
|
Multi-agent AI consultation framework for Claude Code via MCP — second opinions from OpenAI, Gemini, Qwen, DeepSeek. |
Self-hosted runtime for durable AI agents — fractal multi-agent orchestration, 6-layer memory, MCP tools, local dashboard. |
- Context Engineering: How to Manage Context for AI Models and Agents
Practical framework for structuring context — 6 layers, 4 strategies, persistence, and subagents with real Claude Code examples. - Multi-Provider LLM: How to Stop Depending on a Single API
Architect resilient LLM systems with fallback chains, task routing, and cost balancing across providers. - Circuit Breaker in Deno Edge Functions
Build resilience layers for edge functions with circuit breaker patterns, retry logic, and fail-open rate limiting. - LLM Observability with Langfuse
Enable tracing, prompt management, cost tracking, and evaluations with self-hosted Langfuse. - Maestro + Flutter: E2E Tests in YAML Without the Pain
Production patterns for mobile testing — regex matchers, iOS quirks, and AI-powered assertions.
