@equus-ai/sdk

Official JavaScript/TypeScript SDK for the EQUUS AI Infrastructure Platform.

EQUUS provides enterprise-grade AI services:

• 🔍 EmbedCore - Document processing, OCR, and multilingual embeddings
• 📊 RAGGRAPH - Knowledge graph building and semantic querying
• 💬 LMS - RAG-augmented chat with multiple LLM providers
• 🎭 Avatar - AI video avatar presentations and streaming

Installation

npm install @equus-ai/sdk
# or
yarn add @equus-ai/sdk
# or
pnpm add @equus-ai/sdk

Quick Start

Using EQUUS Gateway (Full Platform)

import { Equus } from '@equus-ai/sdk';

const equus = new Equus({
  apiKey: 'eq_live_your_api_key'
});

// Process a document
const doc = await equus.embedcore.process({
  file: myFile,
  eventId: 'event-123'
});

// Search documents
const results = await equus.embedcore.search({
  query: 'keynote speaker schedule',
  eventId: 'event-123'
});

// Chat with RAG
const response = await equus.lms.chat({
  message: 'What are the main topics?',
  eventId: 'event-123'
});

// Start avatar session
const session = await equus.avatar.startSession({
  faceId: 'default',
  voiceId: 'rachel'
});

Using RunPod Direct (GPU-Accelerated Embeddings)

For direct access to GPU-accelerated embedding generation and OCR, bypass the Gateway:

import { EquusRunPod } from '@equus-ai/sdk';

const runpod = new EquusRunPod({
  apiKey: 'your-runpod-api-key',
  embedcoreEndpoint: '7nnw0dqwt1o920' // Default EQUUS endpoint
});

// Warmup models (recommended on first call)
await runpod.warmup();

// Generate embedding - auto-routes based on language
const english = await runpod.embed('Hello world');  // MiniLM (384d)
const hindi = await runpod.embed('नमस्ते दुनिया'); // MuRIL (768d)

// Batch embeddings (up to 100)
const batch = await runpod.embedBatch(['First text', 'Second text']);

// OCR - extract text from image
const ocr = await runpod.ocr(imageBase64);

// OCR - extract product data
const product = await runpod.ocrProduct(productImageBase64);

// OCR - extract menu items
const menu = await runpod.ocrMenu(menuImageBase64);

Services

EmbedCore - Document Processing

// Process document
await equus.embedcore.process({
  file: pdfFile,
  eventId: 'event-123',
  chunkSize: 512
});

// Semantic search
const results = await equus.embedcore.search({
  query: 'What time does registration open?',
  eventId: 'event-123',
  topK: 5
});

// List documents
const docs = await equus.embedcore.listDocuments('event-123');

// Delete document
await equus.embedcore.deleteDocument('doc-id');

RAGGRAPH - Knowledge Graphs

// Build knowledge graph
const graph = await equus.raggraph.build({
  eventId: 'event-123'
});

// Query graph
const answer = await equus.raggraph.query({
  query: 'What are the relationships between speakers?',
  eventId: 'event-123',
  includeContext: true
});

// Get graph stats
const stats = await equus.raggraph.getStats('event-123');

LMS - Chat & RAG

// Simple chat
const response = await equus.lms.chat({
  message: 'Summarize the event',
  eventId: 'event-123'
});

// Streaming chat
for await (const chunk of equus.lms.chatStream({
  message: 'Tell me about the keynote',
  eventId: 'event-123'
})) {
  process.stdout.write(chunk.content || '');
}

// List models
const models = await equus.lms.listModels();

Avatar - AI Video

// Start session
const session = await equus.avatar.startSession({
  faceId: 'face-001',
  voiceId: 'voice-rachel',
  eventId: 'event-123'
});

// Query with video response
const response = await equus.avatar.query({
  sessionId: session.sessionId,
  query: 'What time is lunch?',
  includeVideo: true
});

// Text-to-speech
const speech = await equus.avatar.speak({
  sessionId: session.sessionId,
  text: 'Welcome to the conference!'
});

// Start streaming to Zoom
await equus.avatar.startStream({
  sessionId: session.sessionId,
  platform: 'zoom',
  layout: 'pip'
});

// List faces and voices
const faces = await equus.avatar.listFaces();
const voices = await equus.avatar.listVoices();

// Stop session
await equus.avatar.stopSession(session.sessionId);

Configuration

EQUUS Gateway (Full Platform)

const equus = new Equus({
  apiKey: 'eq_live_your_api_key',  // Required - EQUUS API Key
  baseUrl: 'https://equus-gateway-production.up.railway.app', // Optional
  timeout: 30000,  // Request timeout in ms (default: 30000)
  retries: 3,      // Retry attempts (default: 3)
  debug: true      // Enable debug logging
});

RunPod Direct (GPU Embeddings)

const runpod = new EquusRunPod({
  apiKey: 'rpa_xxx',              // Required - RunPod API Key
  embedcoreEndpoint: '7nnw0dqwt1o920',  // Optional - default endpoint
  timeout: 60000,  // Request timeout in ms (default: 60000)
  debug: true      // Enable debug logging
});

Embedding Model Selection

EQUUS EmbedCore uses a hybrid embedding strategy that automatically routes to the best model based on language detection:

| Model | Dimensions | Use Case | Languages | |-------|------------|----------|-----------| | paraphrase-multilingual-MiniLM-L12-v2 | 384 | General/Default | 50+ languages | | l3cube-pune/indic-sentence-bert-nli (MuRIL) | 768 | Indian Languages | Hindi, Nepali, Tamil, +14 more |

// Auto-detection in action:
const english = await runpod.embed('Hello world');
console.log(english.model_used);      // 'default'
console.log(english.dimension);        // 384

const hindi = await runpod.embed('नमस्ते दुनिया');
console.log(hindi.model_used);         // 'indian'
console.log(hindi.dimension);          // 768

OCR Capabilities

Extract text and structured data from images:

// Basic OCR
const text = await runpod.ocr(imageBase64, ['en', 'hi']);

// Product extraction - returns name, price, SKU, brand
const product = await runpod.ocrProduct(productImage);
console.log(product.name, product.price);

// Menu extraction - returns structured menu items
const menu = await runpod.ocrMenu(menuImage);
menu.items.forEach(item => console.log(item.name, item.price));

Health Checks

// Check gateway health
const health = await equus.health();
console.log(health.status); // 'healthy'

// Check all services
const services = await equus.servicesHealth();
console.log(services.embedcore.status);
console.log(services.raggraph.status);
console.log(services.lms.status);
console.log(services.avatar.status);

// RunPod health
const runpodHealth = await runpod.health();
console.log(runpodHealth.gpu_name);    // 'NVIDIA GeForce RTX 4090'
console.log(runpodHealth.gpu_memory);  // '25.4GB'

Error Handling

try {
  await equus.embedcore.process({ file, eventId });
} catch (error) {
  if (error.status === 401) {
    console.error('Invalid API key');
  } else if (error.status === 429) {
    console.error('Rate limit exceeded');
  } else {
    console.error('Error:', error.message);
  }
}

TypeScript Support

Full TypeScript support with exported types:

import { 
  Equus,
  EquusConfig,
  EquusRunPod,
  RunPodConfig,
  ProcessDocumentRequest,
  ChatRequest,
  ChatResponse,
  SearchResult,
  EmbeddingResult,
  OCRResult,
  // ... more types
} from '@equus-ai/sdk';

Platform URLs

| Service | URL | |---------|-----| | API Gateway | https://equus-gateway-production.up.railway.app | | EmbedCore (RunPod) | https://api.runpod.ai/v2/7nnw0dqwt1o920 | | Documentation | https://equus-gateway-production.up.railway.app/docs |

Requirements

• Node.js >= 16.0.0
• For browser usage: Modern browser with Fetch API support

License

MIT

Support

• GitHub Issues: https://github.com/whatsupdoc-in/equus-platform/issues
• Documentation: https://equus-gateway-production.up.railway.app/docs