Project: Multi-Modal Data Analysis API Description: FastAPI-based REST API that orchestrates LLM workflows for multi-modal data analysis (text, CSV, images, code generation, etc.). This is a generic, shareable README template with no personal data.

Warning: This is a fallback version and not my main version. I may use this in case my version is not able to work or I may use a different fallback altogether.

Note: This work is only a minor modification of https://github.com/MuthuPeru/DataAnalystAgent (This is much better than what I could come up with and is really helping me learn. I really wanted to undertand how we could use Langchain here and despite getting stuck in random errors, I kept trying. Now that I have Muthu's code to learn from, I will continute to work further to enhance my understanding. The original copyright and MIT license belongs to Muthu Peru.)

• REST API built with FastAPI
• Supports multi-file uploads; questions.txt (or similar) is required for analysis requests
• Multiple workflow types (data analysis, image analysis, code generation, EDA, predictive modeling, etc.)
• Synchronous processing for short-running tasks (service-level timeout / limits should be enforced separately)
• Designed to be run locally or in a container

• Multiple file upload support (1 required questions.txt + optional files)
• Intelligent workflow detection (configurable)
• Multi-modal processing (text, images, structured files)
• Enhanced logging and error handling
• Prebuilt workflows for common analytical tasks

• Python 3.10+
• Recommended: virtual environment (venv / conda)
• See requirements.txt or pyproject.toml for full dependency list

Create a .env from the template .env.template and set required values. Example environment variables (do not commit secrets):

• OPENAI_API_KEY — API key for LLM provider (if used)
• LANGCHAIN_TRACING_V2 — enable tracing (optional)
• LANGCHAIN_API_KEY — LangChain tracing key (optional)
• Other provider-specific keys as required by your configuration

• POST /api/ — Submit analysis tasks (multipart/form-data; questions_txt required)
• GET /health — Health check (returns 200 OK if service is healthy)
• GET / — API info / landing page

• POST /api/analyze — JSON-based analysis request
• POST /api/workflow — Execute a specific workflow by name
• POST /api/pipeline — Execute a multi-step pipeline
• GET /api/tasks/{id}/status — Check task status

curl "http://localhost:8000/api/" \
  -F "questions_txt=@questions.txt" \
  -F "files=@dataset.csv" \
  -F "files=@image.png"

import requests

resp = requests.post(
    "http://localhost:8000/api/analyze",
    json={
        "task_description": "Analyze customer churn data",
        "workflow_type": "data_analysis",
        "dataset_info": {
            "description": "Customer dataset",
            "columns": ["age", "tenure", "charges", "churn"],
            "sample_size": 1000
        }
    }
)
print(resp.json())

• data_analysis — General analysis & recommendations
• image_analysis — Image processing / CV pipelines
• text_analysis — NLP & text analytics
• code_generation — Generate executable Python code for analysis
• exploratory_data_analysis — EDA plan & execution
• predictive_modeling — Model guidance & training suggestions
• data_visualization — Chart/plot suggestions or generation
• statistical_analysis — Statistical tests & inference
• web_scraping — Web data extraction guidance
• database_analysis — SQL / DuckDB assistance

1. Clone the repository

git clone <REPO_URL>
cd <REPO_DIR>

2. Create & activate a virtual environment

python -m venv .venv
source .venv/bin/activate   # Linux/macOS
.venv\Scripts\activate      # Windows (PowerShell)

3. Install dependencies

pip install -r requirements.txt

4. Create .env from the template and set variables

cp .env.template .env
# edit .env to add required keys (do not commit)

5. Run the server (development)

uvicorn main:app --reload --host 0.0.0.0 --port 8000

6. Visit the API docs

http://localhost:8000/docs

Build and run using the provided scripts or manually:

# build
docker build -t data-analysis-api .

# run (example)
docker run -d --name data-analysis-api -p 8000:80 --env-file .env data-analysis-api

• Simple tests can be run with included test scripts:

python test_api.py
python test_file_upload_api.py

(Adapt names to your test files)