PamirAI
The BHV guide and GitHub branches document PamirAI's Distiller BHV software surface, while Raspberry Pi's article says PamirAI provided the baseboard and software stack.
SourceBiohacking Village at DEF CON 33 · United States · 2025
Compute Module 5 medical-AI badge
Biohacking Village's DEF CON 33 Distiller BHV badge was a CM5-based medical-AI badge built with SolaSec and PamirAI, combining local voice/chat interaction, e-ink UI, physical buttons, colored LEDs, battery power, and public BHV software branches.
People
The BHV guide and GitHub branches document PamirAI's Distiller BHV software surface, while Raspberry Pi's article says PamirAI provided the baseboard and software stack.
SourceRaspberry Pi's article says SolaSec handled the physical design, 3D-printed enclosures, and final assembly for the Biohacking Village badge.
SourceOfficial Biohacking Village sources publish the DEF CON 33 badge context and medical-chatbot framing used for this record.
SourceIt pushes village badges into portable edge-AI hardware: a full Raspberry Pi Compute Module 5 class board running local medical-assistant workflows rather than a simple microcontroller puzzle, while still fitting Biohacking Village's bio/CTF/event-art tradition.
Public sources document a Raspberry Pi Compute Module 5 core on a custom PCB, an RP2040/Pico board-manager layer, e-ink display drivers, physical buttons, colored LEDs, built-in microphone, SD-card slot, Raspberry Pi GPIO headers, battery setup, USB-C Power Delivery charging requiring 9V/3A, and a 3D-printed enclosure. This pass did not recover a final public schematic, PCB release, bill of materials, or production count.
The public trail describes a local medical-assistant chatbot with voice interaction, onboard LLM behavior, Distiller CM5 SDK and Distiller CM5 Python BHV branches under Apache-2.0 licensing, hardware/audio/camera/e-ink/SAM LED modules, Parakeet ASR/VAD, Piper TTS, optional Whisper, a conversational interface, LLM integration, MCP server support, local GGUF model switching, WiFi setup, debug mode, and SSH access.
The badge was built as a Biohacking Village, SolaSec, and PamirAI collaboration. Raspberry Pi's post-event article says PamirAI supplied the baseboard and software stack while SolaSec handled physical design, 3D-printed enclosures, and final assembly; the same article frames the badge as a private edge-AI medical chatbot and notes Biohacking Village's broader bio-based, technically challenging, CTF-oriented badge tradition.
Lifecycle
The application repository documents a conversational assistant with local model switching, LLM integration, MCP support, hardware interfaces, and a medical-assistant prompt example.
SourceThe BHV guide describes a non-touch e-ink main display, left-side up/down buttons, right-side enter button, and e-ink drivers controlled through the microcontroller and Raspberry Pi 5 side.
SourceThe guide documents an RP2040 microcontroller / Raspberry Pi Pico board manager for button inputs, LED control, battery charging/status, and power management.
SourceRaspberry Pi's article and the PamirAI user guide identify the badge around a Raspberry Pi Compute Module 5 platform intended to run local, private, interactive edge AI.
SourceThe user guide documents battery connection during setup, USB Power Delivery charging with a 9V/3A minimum, SSH access, RGB LED shutdown behavior, and 3D-printing button-tolerance troubleshooting.
SourceThe SDK branch documents hardware, audio, camera, e-ink, SAM LED, ASR/VAD, TTS, optional Whisper, native display, package, and Debian installation modules for the CM5 platform.
SourceOperational history
The record presents the badge as a serviceable field device with operational constraints rather than a frictionless consumer gadget.
The catalogue keeps hardware claims to verified public documentation and avoids inventing component identifiers or manufacturing details.
The badge is recorded as an event artifact and AI interaction surface, not as a trusted medical device or medical-advice system.
The record remains source-backed and image-free rather than copying page media, documentation screenshots, article photos, social posts, or generated imagery.