Maker Builds Offline AI-Powered BMO Replica
Maker Brenpoly built a functional, offline BMO replica using a Raspberry Pi 5 (16GB), Raspberry Pi Camera Module V2, a Pimoroni NVMe Base Duo, Geekworm UPS HAT, 5” touchscreen, lithium battery, USB microphone and speaker, and Adafruit Feather 32u4 buttons. He runs local models with Ollama — Gemma 3 for dialogue and Moondream 2 for vision — using Whisper for audio input and Piper TTS for speech output. The build demonstrates on-device multimodal AI for embodied assistants.
Key Points
- 1Builds a fully offline BMO replica using Raspberry Pi 5, custom PCB, 3D-printed enclosure.
- 2Uses local Gemma 3 LLM and Moondream 2 VLM via Ollama for on-device multimodal intelligence.
- 3Enables practitioners to deploy embodied, offline AI agents on constrained edge hardware with privacy benefits.
Scoring Rationale
Practical demonstration of offline LLM/VLM on edge hardware, limited by single-maker detail and shallow technical depth.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
