TuneTask – AI-Powered Task Manager

TuneTask is a Task Manager that leverages AI (OpenAI GPT-4) for intelligent semantic search of tasks. It is built with .NET 8, ADO.NET, JWT Authentication, and Docker for scalability and performance.

---

🚀 Features

✅ Core Features

• 🗑️ Task Management – Create, update, and retrieve tasks.
• 🔐 Authentication & Authorization – Secure API with JWT tokens.
• 🌜 Error Handling & Logging – Global exception handling with structured logging.
• 🐳 Dockerized SQL Database – Runs in a containerized SQL Server instance.
• 🔍 Semantic Serach with AI - The solution uses OpenAI's semantic search to provide best based not only on exact term matches but similarity.

⚙️ Tech Stack

Technology	Purpose
.NET 8	Backend API
ADO.NET	Direct SQL interactions
JWT Authentication	Secure user authentication
AI (OpenAI API)	AI-powered search
Docker & Docker Compose	Containerized database
MSSQL (via Docker)	Database

---

🔠 Architecture Overview

TuneTask follows Clean Architecture principles:

/src
  /TuneTask.Core         # Business Logic & Interfaces
  /TuneTask.Infrastructure  # Data Access (ADO.NET)
  /TuneTask.Shared       # Error Handling 
  /TuneTask.API          # REST API (Controllers & Authentication)
  /TuneTask.Test          # Unit Testing of solution

---

🛠️ Setup & Installation

1️⃣ Clone the Repository

git clone https://github.com/estebankt/TuneTask.git
cd TuneTask

2️⃣ Configure OpenAI API Key

Replace the token in appsettings.json in the TuneTask.Api project:

"OpenAI": {
  "ApiKey": "YOUR_OPENAI_API_KEY"
}

3️⃣ Run the Application with Docker

run docker compose from Visual Studio

4️⃣ Access the API

• Swagger UI: http://localhost:5051/swagger
• API Endpoints: Endpoints are explained below

---

🛡️ Security & Authentication

• JWT Authentication:
  • Register: POST /api/auth/register
  • Login: POST /api/auth/login → Receive JWT Token
  • Protected Endpoints require Authorization: Bearer {token} header.

---

📺 AI-Powered Tasks Search Instructions

🚀 How AI-Powered Search Works

TuneTask leverages OpenAI's text embedding model (text-embedding-3-small) to enable semantic search. This allows users to search tasks not just by keywords but also by meaning, improving search relevance.

🔎 How It Works:

1. Task Creation with Embeddings:

   • When a task is created, its description is processed using OpenAI’s embedding model.
   • The model converts the text into a high-dimensional vector representation (embedding).
   • This vector is stored in the database along with the task.

2. Search Query Processing:

   • When a user searches for a task, the query text is also converted into an embedding using the same model.
   • Instead of a direct SQL LIKE query, we compare the stored task embeddings with the query embedding using cosine similarity.

3. Retrieving Relevant Tasks:

   • The search function retrieves all tasks from the database.
   • It computes the cosine similarity between the query embedding and stored embeddings.
   • The most relevant tasks (tasks with the highest similarity score) are returned.

🔢 Example:

Scenario:

• User Query: "Work on deep learning project"
• Stored Tasks:
  1. "Train AI Model for NLP"
  2. "Fix UI bugs in React App"
  3. "Read Deep Learning Papers"

Expected Result:

Instead of returning tasks based on exact word matches, the system recognizes that "Train AI Model for NLP" is related to "Work on deep learning project" and ranks it higher.

---

⚙️ Implementation Details

• Model Used: text-embedding-3-small
• Vector Storage: Stored as a FLOAT[] array in the database.
• Similarity Calculation: Cosine Similarity function measures how close the query is to stored task descriptions.

---

🛠️ Using the AI Search API

• Endpoint:

  GET /api/tasks/search?query=your_text_here

• Example Request:

  curl -X GET "http://localhost:5050/api/tasks/search?query=deep+learning" -H "Authorization: Bearer YOUR_TOKEN"

• Example Response:

  [
    {
      "id": "3585e29f-b2b0-4d2d-bbb6-22bdfdb2a762",
      "title": "Train AI Model for NLP",
      "description": "Developing a machine learning model for NLP tasks.",
      "status": "Pending",
      "createdAt": "2025-02-16T15:04:33.1033352Z",
  
    }
  ]

This allows intelligent task retrieval based on meaning rather than just keywords.

---

🚀 API Usage Guide

2️⃣ User Authentication

🔹 Register a New User

Request:

{
  "username": "admin1232",
  "email": "testuser3@example.com",
  "password": "password",
  "role": "Admin"
}

Response:

200 OK
{
  "message": "User registered successfully."
}

🔹 Login to Get JWT Token

Request:

{
 "email": "testuser3@example.com",
  "password": "password"
}

Response:

{
  "token": "<your-jwt-token>"
}

Copy the token value and authenticate with Bearer + Token in the Authorization header.

---

3️⃣ Task Management

🔹 Get All Tasks (Admin Only)

Endpoint:

GET /api/tasks/all

🔹 Get Task by ID

Endpoint:

GET /api/tasks/{taskId}

Example Request:

GET /api/tasks/a5cc4cc0-6084-445e-8874-17ebe5e52da2

Response:

{
  "id": "a5cc4cc0-6084-445e-8874-17ebe5e52da2",
  "userId": "21cc8a68-aec8-402b-9864-179300f491d3",
  "title": "Deep Focus Coding",
  "description": "A long coding session with no distractions.",
  "createdAt": "2025-02-16T16:18:40.207",
  "status": 0,
  "embedding": null
}

---

4️⃣ Create a Task

Endpoint:

POST /api/tasks/create

Example Request:

{
  "title": "Documentation",
  "description": "Document something about a book Old Man and the Sea"
}

Response:

{
  "id": "e299ccae-ef53-4705-8e13-149be59e55da",
  "userId": "e9fe53b5-ee15-4a23-9187-fa96578635c2",
  "title": "Setting up dev environment",
  "description": "Clone repository and have it running on local machine",
  "createdAt": "2025-02-16T20:11:52.4851879Z",
  "status": 0,
  "embedding": [ ... , .. , .., ]
}

---

5️⃣ Update a Task

Endpoint:

PUT /api/tasks/update/{taskId}

Example Request:

{
  "id": "e299ccae-ef53-4705-8e13-149be59e55da",
  "description": "changed"
}

Response:

• 204 No Content (Success)
• "Task not found." (If Task ID does not exist)

---

6️⃣ Delete a Task

Endpoint:

DELETE /api/tasks/delete/{taskId}

Response:

• 204 No Content (Success)
• "Task not found." (If Task ID does not exist)

---

7️⃣ AI-Powered Task Search

Endpoint:

GET /api/tasks/search?query=<search-term>

Example:

GET /api/tasks/search?query=focus

Response: (Ordered by best match)

[
  {
    "id": "abc123",
    "title": "Deep Focus Coding",
    "description": "A long coding session with no distractions.",
    "createdAt": "2025-02-16T16:18:40.207",
    "status": 0
  },
  {
    "id": "xyz456",
    "title": "Focused Writing",
    "description": "Writing session with noise-canceling headphones.",
    "createdAt": "2025-02-16T18:22:10.119",
    "status": 0
  }
]

📜 License

This project is licensed under the MIT License.