Core Technologies
This page describes the technology stack and key dependencies that power the GamiBot platform.
Technology Stack
| Component | Technology | Purpose |
|---|---|---|
| LMS | Moodle (custom PHP plugins) | Course and resource management |
| Orchestration | LangFlow | Workflow design and agent orchestration |
| Vector Database | Qdrant | Semantic search and vector storage |
| Embeddings | Configurable (OpenAI, local, etc.) | Convert text to dense vectors |
| LLM | GPT-4, Claude 3, or local alternatives | Agent reasoning and generation |
| ORM | Prisma | Database abstraction (if using relational DB) |
| Backend | Node.js + Express or Python | API layer and webhook handlers |
| File Processing | Unstructured, PyPDF2, python-pptx | Extract text from diverse formats |
Key Dependencies
Moodle Plugins
local_gamibot_manager- Webhook dispatcher for file upload events
LangFlow Integration
- Bundled Qdrant integration for vector operations
- LLM connectors for multiple providers (OpenAI, Anthropic, local models)
Vector Operations
- Qdrant client library - Python or Node.js SDK for vector operations
- Collections support sharding and replication for scalability
Text Extraction Libraries
| Format | Library | Notes |
|---|---|---|
| PyPDF2, pdfplumber | OCR fallback available | |
| PowerPoint | python-pptx | Extracts slides + speaker notes |
| ePub | ebooklib | Chapter parsing support |
| Word | python-docx | Full document support |
| Plain Text | Native | UTF-8 encoding |
Embedding Models
The platform supports configurable embedding models:
OpenAI Embeddings
json
{
"model": "text-embedding-3-small",
"dimensions": 1536,
"max_tokens": 8191
}Local Alternatives
- Sentence Transformers - Self-hosted option for privacy
- all-MiniLM-L6-v2 - Lightweight, efficient embeddings
- BGE - Multilingual support
LLM Provider Options
Cloud Providers
| Provider | Model | Best For |
|---|---|---|
| OpenAI | GPT-4, GPT-4o | General purpose, high quality |
| Anthropic | Claude 3 Opus/Sonnet | Long context, safety |
| Gemini | Multimodal, cost-effective |
Local/Self-Hosted
| Model | Framework | Notes |
|---|---|---|
| Llama 3 | Ollama, vLLM | Open source, privacy |
| Mistral | Ollama, vLLM | Efficient, multilingual |
| Phi-3 | Ollama | Small, fast |
Infrastructure Requirements
See the Deployment Architecture page for detailed infrastructure specifications.