docs: jaqpot docker models section (#40)

docusaurus/docs/docker-models/deploy-your-model.md

@@ -0,0 +1,51 @@
+---
+sidebar_position: 3
+---
+
+# Deployment
+
+This guide explains how to deploy your Docker model to Jaqpot.
+
+## Creating the model configuration
+
+First, you need to register your model with Jaqpot by sending a configuration to the `/v1/models` endpoint:
+
+```json
+{
+  "name": "Your model name",
+  "type": "DOCKER",
+  "description": "model description",
+  "independentFeatures": [
+    {
+      "key": "input",
+      "name": "Input",
+      "featureType": "TEXT"
+    }
+  ],
+  "dependentFeatures": [
+    {
+      "key": "output",
+      "name": "Output",
+      "featureType": "TEXT"
+    }
+  ],
+  "task": "BINARY_CLASSIFICATION",
+  "visibility": "PUBLIC",
+  "dockerConfig": {
+    "appName": "your-app-name" 
+  }
+}
+```
+
+The `appName` in the configuration is a unique identifier for your model in the Jaqpot system.
+
+## Deploying your Docker image
+
+After registering your model configuration:
+
+1. Build your Docker image using the provided Dockerfile
+2. Push the image to a Docker registry
+3. Share your Docker image name with us
+4. We will host your image and connect it to Jaqpot using your specified `appName`
+
+Once deployed, your model will be available through the standard Jaqpot prediction endpoints.

docusaurus/docs/docker-models/implementing-your-model.md

@@ -0,0 +1,78 @@
+---
+sidebar_position: 2
+---
+
+# Implementing your model
+
+This guide explains how to implement your custom model for Jaqpot using Docker.
+
+## Getting started
+
+First, fork our template repository: https://github.com/ntua-unit-of-control-and-informatics/jaqpot-model-with-docker-template
+
+The template includes:
+- A training script (`train.py`)
+- A FastAPI server implementation (`main.py`)
+- Prediction request/response models
+- A Dockerfile for deployment
+
+## Training your model
+
+The `train.py` file is where you implement your model's training logic:
+
+```python
+# train.py
+import pandas as pd
+from sklearn.datasets import make_classification
+from sklearn.ensemble import RandomForestClassifier
+import joblib
+
+# Generate sample data
+X, y = make_classification(n_samples=1000, n_features=4, n_informative=2, n_redundant=0)
+
+df = pd.DataFrame(X, columns=["X1", "X2", "X3", "X4"])
+
+# Train a sample model
+model = RandomForestClassifier(n_estimators=100)
+model.fit(df, y)
+
+# Save the model
+joblib.dump(model, 'model.pkl')
+```
+
+This creates a `model.pkl` file that will be loaded by your prediction service.
+
+## Implementing the prediction endpoint
+
+Your main task is implementing the `/infer` endpoint. Here's a basic example:
+
+```python
+from fastapi import FastAPI
+import pickle
+from app.models import PredictionRequest, PredictionResponse
+
+app = FastAPI()
+
+# Load the trained model
+with open('model.pkl', 'rb') as f:
+    model = pickle.load(f)
+
+@app.post("/infer", response_model=PredictionResponse)
+async def predict(request: PredictionRequest):
+    result = model.predict(request.data)
+    return PredictionResponse(status=200, data=result)
+```
+
+## Testing locally
+
+1. Train your model:
+```bash
+python train.py
+```
+
+2. Run the FastAPI server:
+```bash
+uvicorn app.main:app --reload
+```
+
+Your model is now ready for Docker deployment on Jaqpot.

docusaurus/docs/docker-models/index.mdx

@@ -0,0 +1,29 @@
+---
+sidebar_position: 5
+title: Docker models
+---
+
+import DocCardList from '@theme/DocCardList';
+
+# Docker models on Jaqpot
+
+Jaqpot supports deploying custom models using Docker containers. This lets you run any model, regardless of its framework or dependencies, as long as you package it correctly.
+
+### Important note about deployment
+
+The deployment process for Docker models is manually managed by our team to ensure security and proper integration. Before deployment:
+
+1. Contact us at [email protected] with your model details
+2. We will review your requirements and provide specific instructions
+3. Once approved, we will help you integrate your model into our infrastructure
+
+### How it works:
+
+1. Create a model configuration with Docker settings
+2. Implement your model using our template
+3. Build and test your Docker image locally
+4. Contact us for deployment review and instructions
+5. After approval, we'll host your model on Jaqpot
+
+The following sections will guide you through the technical implementation. Remember to reach out to us before starting the deployment process.
+<DocCardList />

docusaurus/docusaurus.config.ts

@@ -150,10 +150,10 @@ const config: Config = {
     },
     algolia: {
       // The application ID provided by Algolia
-      appId: process.env.ALGOLIA_APP_ID,
+      appId: process.env.ALGOLIA_APP_ID || 'local',
 
       // Public API key: it is safe to commit it
-      apiKey: process.env.ALGOLIA_SEARCH_API_KEY,
+      apiKey: process.env.ALGOLIA_SEARCH_API_KEY || 'local',
 
       indexName: 'jaqpot',