Cloud-dependent smart home platforms introduce critical limitations including high command execution latencies, complete functionality loss during network outages, security vulnerabilities, and inadequate linguistic support for morphologically rich languages like Ukrainian. This research addresses these challenges by developing a hybrid local-cloud architecture that maintains device control autonomy while preserving user interface accessibility. The proposed system integrates containerized Home Assistant deployment, MQTT message broker for local communication, and spaCy-based natural language processing with Pymorphy3 morphological analysis specifically adapted for Ukrainian language commands. The methodology employs lemmatization and tokenization to handle complex grammatical variations including case systems, gender agreements, and aspectual distinctions characteristic of Ukrainian morphology. Experimental validation demonstrates high intent recognition accuracy across single-device, multi-device, and conditional command categories while maintaining sub-second response times on consumer-grade hardware. The system achieves complete operational independence during internet disruptions and ensures zero external data transmission through fully local processing architecture. Theoretical contributions include establishing first documented benchmarks for Ukrainian language intent recognition in smart home contexts, demonstrating sub-linear computational complexity for edge-based processing, and validating hybrid architectural models balancing sovereignty with accessibility. Practical implications encompass complete data sovereignty, elimination of subscription dependencies, vendor independence through open standards, and community-extensible open-source implementation. The research establishes reproducible methodology applicable to other underserved languages and resource-constrained IoT environments while addressing significant market gaps in Ukrainian-speaking populations.