System Requirements Document for mango-ai

1. Introduction

The mango-ai project is an AI-driven supply-chain risk prediction tool designed to assist supply-chain risk analysts in predicting and assessing potential risks using advanced AI models. The tool leverages Python, XGBoost, time-series methods, AWS, and FastAPI to deliver accurate and timely predictions, providing analysts with the insights needed to make informed decisions.

2. System Overview

The mango-ai system is designed to predict supply-chain risks using AI models. It is built on a stack that includes Python, XGBoost, time-series analysis, AWS, and FastAPI. The system is intended for use by supply-chain risk analysts who will interact with the tool to review predictions and assess risks. The system includes several custom pages for user interaction, each serving a specific part of the prediction and assessment lifecycle.

2a. Product Interpretation and Delivery Boundary

The mango-ai system is delivered as a web application with custom UI components. It requires analysts to self-enroll and verify their identity to access protected prediction and assessment functionalities. The system's backend is responsible for handling data storage, prediction runs, and model outputs. The application is designed to be accessed via a secure login, ensuring that only authorized analysts can access sensitive prediction data.

2b. Source Content Inventory

Not applicable as no content_source directive was provided.

2c. Page Content and Component Coverage

Landing

• Purpose: Introduce the AI supply-chain risk prediction product and its capabilities.
• Components:
  • Overview of the system and its benefits.
  • Explanation of the prediction workflow.
  • Call-to-action for analysts to enroll or log in.

Login

• Purpose: Allow analysts to verify their identity and access protected features.
• Components:
  • Username and password fields.
  • "Forgot Password" link.
  • Login button.

Dashboard

• Purpose: Provide a summary of prediction activities and access to other functionalities.
• Components:
  • Summary of recent prediction runs.
  • Navigation to Data & Runs, Predictions, and Risk Assessment pages.
  • User profile and settings access.

Data & Runs

• Purpose: Manage data inputs and initiate or monitor prediction runs.
• Components:
  • Data upload and management interface.
  • Initiate new prediction runs.
  • Monitor ongoing and past runs.

Predictions

• Purpose: Display current and historical prediction results.
• Components:
  • List of prediction results with timestamps.
  • Filters for sorting and searching predictions.
  • Detailed view of individual prediction results.

Risk Assessment

• Purpose: Review and assess prediction results in the context of supply-chain risks.
• Components:
  • Tools for analyzing prediction data.
  • Risk assessment reports.
  • Export options for assessment data.

3. Functional Requirements

• As a Supply-Chain Risk Analyst, I should be able to enroll in the system to access prediction functionalities. (required_inference)
• As a Supply-Chain Risk Analyst, I should be able to log in to resume my prediction and assessment work. (required_inference)
• As a Supply-Chain Risk Analyst, I should be able to upload data and initiate prediction runs. (explicit)
• As a Supply-Chain Risk Analyst, I should be able to view and analyze prediction results. (explicit)
• As a Supply-Chain Risk Analyst, I should be able to assess risks based on prediction results. (explicit)

4. User Personas

• Supply-Chain Risk Analyst: Responsible for reviewing AI-generated supply-chain risk predictions and using them for risk assessment. The analyst interacts with all system pages to manage data, run predictions, and assess risks.

5. Core User Flows

1. Enrollment and Login Flow:

   • The analyst visits the Landing page and chooses to enroll or log in.
   • Upon choosing to enroll, the analyst provides necessary information and creates an account.
   • The analyst logs in using their credentials to access the Dashboard.

2. Data Management and Prediction Flow:

   • From the Dashboard, the analyst navigates to the Data & Runs page.
   • The analyst uploads data and initiates a prediction run.
   • The system processes the data and updates the analyst on the run's status.

3. Prediction Analysis Flow:

   • The analyst accesses the Predictions page from the Dashboard.
   • The analyst reviews prediction results, using filters and search tools as needed.
   • The analyst selects a prediction to view detailed results.

4. Risk Assessment Flow:

   • The analyst navigates to the Risk Assessment page from the Dashboard.
   • The analyst uses available tools to assess risks based on prediction results.
   • The analyst generates and exports risk assessment reports.

6. Visuals Colors and Theme

• Muse: Gleb Kuznetsov
• Palette:
  • Background: #000814
  • Surface: #001D3D
  • Text: #E0E1DD
  • Primary: #0077FF
  • Accent: #FF00FF
  • Muted: #4B4B4B
• Typography:
  • Headings: Space Grotesk, Bold, uppercase, tight tracking
  • Body: Orbitron
  • Scale: 1.25 modular, 60/48/34/24/18
• Shape Language: Floating glass panels, thin luminous strokes, radial layout
• Layout: Full-bleed 3D scene with floating panels, HUD-like data presentation

7. Signature Design Concept

The Landing page will feature a full-bleed 3D scene with floating glass panels overlaid with dynamic data streams. Electric cyan glows will highlight interactive elements, creating an immersive experience that emphasizes the tool's advanced AI capabilities.

8. Interaction Model & Motion Direction

• Interaction Model: Animated
• Motion Tempo: Cinematic
• Hero Dimensionality: WebGL
• Landing Hero Motion Brief: The hero section will feature a continuous slow orbit of abstract 3D objects with parallax effects. Data streams will animate across floating glass panels, creating a dynamic and immersive entry experience.

9. Non-Functional Requirements

• The system must be hosted on AWS to ensure scalability and reliability. (explicit)
• The system must use FastAPI for efficient backend processing. (explicit)
• The system must handle time-series data efficiently to support prediction accuracy. (explicit)

10. Tech Stack

• Programming Language: Python
• Machine Learning: XGBoost
• Data Processing: Time Series
• Infrastructure: AWS
• API Framework: FastAPI

11. Assumptions and Constraints

• The system assumes that analysts have basic knowledge of supply-chain risk management.
• The system is constrained to use the specified tech stack: Python, XGBoost, Time Series, AWS, and FastAPI.

12. Glossary

• Supply-Chain Risk Analyst: A professional responsible for evaluating risks in the supply chain using predictive analytics.
• Prediction Run: The process of executing the AI model to generate risk predictions based on input data.
• Risk Assessment: The evaluation of prediction results to determine potential risks in the supply chain.