Homy Robotics

What I built

AI Engineering Intern (April – June 2026). Built the full voice stack — real-time audio, swappable STT, LLM, TTS, and a 30-table backend — and took a robot that had never spoken to its first working deployment.

Fixed a 30-minute install on low-memory hardware (cut to 2 minutes) and an audio routing bug that sent output to a non-existent device. Added household-isolated memory with semantic recall and verbal bridging so conversation never stalls during retrieval.