Kyrylo Lukianenko

At Rentals United, I designed and built backend microservices for a vacation-rental channel manager, connecting platforms including Airbnb, Booking.com and Expedia. I rewrote hot-path queries from EF Core to Dapper, cutting availability API p95 latency from 800 ms to 120 ms at ~1M requests per day.

I also built real-time updates over WebSockets and SignalR, plus queue-based synchronization with AWS SQS. Multi-level caching and retry policies cut average response time on frequently used endpoints by ~40% and failed sync calls by ~30%.

In my independent product work, I’ve shipped five LLM features into production. I’ve built .NET and Node.js agents using Anthropic Claude and OpenAI APIs, including a tool-using agent with MCP servers and a scheduled job-search agent that filters and deduplicates listings before LLM scoring.

For MyFamily, I built a voice assistant that retrieves family data through tools and returns changes validated against a strict JSON Schema. I also created evaluation cases to compare models and settings, and built vision and transcription pipelines for receipts and documents.