NOESIS

---

Overview

Noesis is an AI-powered data analysis and processing platform designed to streamline the workflow of data scientists and analysts. By leveraging the power of Large Language Models (LLMs) and the Model Context Protocol (MCP), Noesis automates complex data tasks ranging from exploratory data analysis to advanced preprocessing and visualization.

Noesis provides a unified interface for interacting with datasets, enabling users to perform sophisticated data operations through Natural Language Commands or structured workflows.

---

Demo Video

NoesisDemoVideo.mp4

---

Architecture

The system follows a Client-Server Architecture where the frontend communicates with the FastAPI backend. The backend orchestrates data operations by delegating tasks to specialized AI agents and MCP tools. Data persistence is managed through PostgreSQL for application state and MinIO for large datasets and file artifacts.

Reasons for this Architecture:

• Scalability: The frontend and backend can be scaled independently based on demand.
• Separation of Concerns: Decoupling the user interface from the business logic ensures cleaner code and easier maintenance.
• Flexibility: Allows for the use of specialized technologies (Next.js for UI, Python/FastAPI for AI and data processing) where they excel.
• Performance: Next.js provides a fast, optimized user experience, while FastAPI ensures high-performance API handling.

---

Key Features

Exploratory Data Analysis (EDA)

Gain immediate insights into your datasets with automated analysis tools.

• Dataset Summarization: Automatically generate descriptive statistics to understand data distribution.
• Outlier Detection: Identify anomalies using statistical methods like Z-score and IQR.
• Correlation Analysis: Generate correlation matrices to discover relationships between variables.

Data Preprocessing

Clean and prepare your data for modeling with a suite of transformation tools.

• Data Cleaning: Remove duplicates and handle missing values with configurable strategies.
• Type Conversion: Intelligent data type detection and conversion.
• Feature Engineering: Encode categorical variables and normalize numerical columns.
• Outlier Handling: Automatically treat or remove detected outliers.

Data Visualization

Visualize your data with high-quality, interactive charts.

• Statistical Plots: Generate histograms, box plots, and scatter plots to analyze distributions and relationships.
• Categorical Analysis: Visualize categorical data with bar charts.
• Trend Analysis: Track changes over time with line charts.
• Heatmaps: Visualize correlation matrices for quick pattern recognition.

---

Tech Stack

Noesis is built on a modern, scalable architecture utilizing industry-standard technologies.

AI & Data Engine

• Orchestration: LangChain and LangGraph for managing complex AI agent workflows.
• Observability: LangSmith for tracing, monitoring, and debugging LLM applications.
• Protocol: Model Context Protocol (MCP) for standardized tool interaction.
• Data Processing: Pandas and Scikit-learn for efficient data manipulation and analysis.
• Visualization: Matplotlib and Seaborn for generating static and dynamic plots.

Backend

• Framework: FastAPI (Python) for high-performance API endpoints.
• Database: PostgreSQL for structured data storage.
• Object Storage: MinIO for scalable file and dataset storage.
• Authentication: OAuth2 with JWT tokens for secure access control.

Frontend

• Framework: Next.js 15 (React 19) for a responsive and server-side rendered user interface.
• Language: TypeScript for type-safe code.
• Styling: Tailwind CSS v4 for modern, utility-first design.
• State Management: Zustand for efficient client-side state handling.

---

If you have any query, feedback or suggestion feel free to drop a mail at chetan.mahale0220@gmail.com :)