Shreyash AI Assistant Local-MCP-Ollama-Setup

This guide explains how to set up a local AI Assistant using:

• Ollama
• Visual Studio Code
• Postman
• Python backend (MCP Server)

Everything runs on a local machine.

---

1. Install Ollama

Download and install: https://ollama.com

ollama --version  # Verify installation

---

2. Download a Model

Example model: https://ollama.com/library/qwen2.5:7b-instruct

ollama pull qwen2.5:7b-instruct  # Pull the model

ollama run qwen2.5:7b-instruct  # Run the model

If it responds in terminal, setup is correct.

---

3. Model Location (Mac)

/Users/shreyash/.ollama/models   # Default model path on my mac | check model location in Ollama settings.

---

4. Test Ollama via API

Test using Postman: Ollama DOC

POST

http://localhost:11434/api/chat

Body → JSON:

{
  "model": "qwen2.5:7b-instruct",
  "messages": [
      {"role": "system","content": "You are helpfull AI assistant"},
      {"role": "user","content": "Explain the reflection"}
  ],
  "stream": false
}

---

5. Create Two Projects

Create two empty projects:

• mcp-server
• Ai-Assistant

Open both in separate Visual Studio Code windows.

---

6. Create Virtual Environment (Python 3.13)

Inside each project:

python3.13 -m venv .venv

Activate virtual environment (Mac/Linux):

source .venv/bin/activate

---

Add the following section after the virtual environment step in your README.md.

---

7. Install Required Dependencies

Create a requirements.txt file inside both projects

Add the required dependencies in requirements.txt

pip install -r requirements.txt   # Install dependencies:

Verify installation:

pip list  #Now your environment is ready to build the MCP server and backend API.

---

8. Create main.py in AI Assistant project and Run Server

Create a main.py file inside your project.

Run the server:

UVICORN_HOST=127.0.0.1 UVICORN_PORT=9001 uvicorn main:app --reload

---

9. Setup MCP Server

Create a new empty project for the MCP server and open it in Visual Studio Code.

Create a virtual environment (if not already created):

python3.13 -m venv .venv
source .venv/bin/activate

Create a requirements.txt file and add the dependencies mentioned in Step 7.

Install dependencies:

pip install -r requirements.txt

Copy the main.py file into this MCP project.
(Implementation details will be added in the repository.)

---

Install uv

Install uv using terminal:

pip install uv

Verify installation:

uv --version

---

Run MCP Server

uv run main.py

---

Add the following correction under your AI Assistant server section.

---

Test AI Assistant API (Postman)

http://127.0.0.1:9001/chat

Body → JSON:

{
  "query": "What is the current bitcoin price"
}

---

This repository is for learning and personal experimentation with local AI systems and MCP architecture.