Project Cynosura

<languages /> <translate>

Project Cynosura is a self-hosted, edge-optimized educational framework and hardware/software pipeline designed for structured 1-hour bidirectional language exchange over video calls.

The system is engineered specifically for asymmetrical language proficiency levels (Beginner Vietnamese / Advanced Conversational English) and constrained over-the-road (OTR) mobile operating environments using VR edge display nodes, self-hosted sync engines, and custom MediaWiki tracking templates.

0. Phase 0: Minimal Viable Setup (KISS Protocol)

Before scaling to cloud-hosted containers and spatial VR setups, initial calls use off-the-shelf mobile redundancy:

• Video Stream: Google Meet / FaceTime over mobile hotspot.
• Asymmetrical Closed Captions:
  • Student (Isidore): Runs Google Translate (VN -> EN) locally on iPhone.
  • Instructor (Linh): Runs Google Translate (EN -> VN) locally on her device.
• Live Shared Workspace: Single shared Google Doc for real-time vocabulary and idiom typing.
• Flashcard Logging: Manual entry into Anki / AnkiWeb during post-call review.

1. Core Principles & Session Framing

To eliminate "language bleeding" (where the dominant shared language takes over conversation), calls operate under strict time-boxing:

Block	Time	Focus Area	Target Output
Block 1: Vietnamese	30 Mins	Tone contour accuracy, S-V-O mechanics, pitch shadow drills	Master 5 survival phrases & 6 tone contours
Block 2: English	30 Mins	Idiom deconstruction, register/context selection, nuance	Practice 3 target idioms across 3 conversational registers

30-Minute Internal Lifecycle

• 00:00–05:00 – Review & Anki deck audit (3–5 past items).
• 05:00–20:00 – Core topic output & pitch contour / idiom mapping.
• 20:00–30:00 – Guided roleplay with real-time, asynchronous wiki logging.

2. Technical Architecture & Stack

The system decouples compute heavy-lifting from mobile endpoints by hosting the desktop workspace on cloud infrastructure while leveraging a Meta Quest 2 inside the cab for multi-monitor spatial rendering.

+-----------------------------------------------------------------------+
|                            TRUCK CAB EDGE                             |
|                                                                       |
|  [ iPhone ] --------(5G Hotspot / USB Gateway)                        |
|      |                                   |                            |
|  [ Quest 2 ] <---(Wi-Fi / Immersed)---> [ Raspberry Pi Node ]        |
|   (VR Headset)                          (Local Offline Buffer/Router) |
+-----------------------------------------------------------------------+
                                   |
                          (WireGuard Tunnel)
                                   v
+-----------------------------------------------------------------------+
|                     DIGITALOCEAN CLOUD CORE                           |
|                                                                       |
|  +------------------------+      +---------------------------------+  |
|  |     MediaWiki Box      |      |     Kasm / XRDP Web Desktop     |  |
|  | (Curriculum Engine +   |      |  (Hosts Whiteboard, Video Call, |  |
|  |  Anki Sync Server)     |      |   & Shared Editing Workspace)   |  |
|  +------------------------+      +---------------------------------+  |
+-----------------------------------------------------------------------+

Software Stack

• Knowledge Base & Curriculum Engine: Self-hosted MediaWiki.
• Flashcard Synchronization: Containerized anki-sync-server Docker instance running alongside MediaWiki.
• Remote Workspace Display: DigitalOcean Droplet running Kasm Web / XFCE streaming to Immersed VR on the Quest 2.
• Interactive Whiteboard: Excalidraw / Mermaid extension embedded into wiki pages.

Hardware Stack

• Primary Node: Meta Quest 2 Headset (virtual multi-display spatial terminal).
• Gateway Node: iPhone (5G tethering & backup client).
• Auxiliary Node (Optional): Raspberry Pi 4/5 + OLED Touchscreen + Rii i8+ Handheld Keyboard running kiwix-serve / local MediaWiki offline buffer.
• Audio: Closed-back headset microphone directly into Quest 2 for cab ambient noise cancellation.

3. Curriculum Structure & Namespaces

The system divides tracking into two dedicated namespaces on this wiki:

• Vietnamese: Focuses on tone mapping ($Ngang, Huyền, Hỏi, Ngã, Sắc, Nặng$), Southern vs. Northern dialect markers, and functional vocabulary.
• English: Focuses on idiom deconstruction matrices, tone of voice, formal vs. casual context, and American cultural idioms.

Structural Template: `Template:LessonBlock`

All individual lesson entries are generated using the following standard template call:

{{LessonBlock
| Session_ID        = 001
| Date              = {{CURRENTYEAR}}-{{CURRENTMONTH}}-{{CURRENTDAY2}}
| Instructor        = Linh
| Student           = Isidore
| Target_Language   = Vietnamese
| Focus_Tones       = Hoi / Nga
| Idioms_Reviewed   = Hit the nail on the head; Under the weather
| Audio_Log         = File:VN_Session_001_Tones.ogg
| Notes             = Real-time corrections logged here
}}

4. OTR Operating Procedure

1. Pre-Call (T-5 min): Park vehicle, establish 5G hotspot. Boot Quest 2, launch Immersed, connect to DigitalOcean desktop droplet.
2. Display Layout (Virtual Space):
   1. Screen 1 (Left): Video Conference (Google Meet / Zoom).
   2. Screen 2 (Center): MediaWiki LessonBlock workspace for real-time async note logging.
   3. Screen 3 (Right): Shared Anki Web / Drawing canvas for tone contours.
3. Post-Call (T+5 min): Export audio clips recorded during the session, drop onto MediaWiki, and run Anki deck sync.

5. System Roadmap

• [x] Program conceptualization & schedule design.
• [ ] Spin up anki-sync-server Docker container on DigitalOcean.
• [ ] Deploy Template:LessonBlock and namespaces on MediaWiki.
• [ ] Test Immersed latency over cellular hotspot in cab environment.
• [ ] (Optional) Flash Raspberry Pi image with offline local buffer mirror.

```

--- </translate>