LLM Emisisons Tracker

This project aims create a standardized way of tracking, storing and retreiving LLM emissions data.

Getting Started

Prerequisites

• Node.js (v20+) and npm
• Java (JDK 21)
• Python (3.13+) and uv - Install uv
• Docker and Docker Compose

Quick Start

# Clone the repository
git clone https://github.com/harveymmaunders/llm-emissions-tracker.git
cd llm-emissions-tracker

# Install Nx workspace
npm install

# Install project dependencies (optional - install only what you need)
npx nx install llm-emissions-tracker-sdk-javascript
npx nx sync llm-emissions-tracker-sdk-python

# Or install all dependencies
npm run install-all

# Build everything
npx nx run-many -t build

# Start database and API
npx nx up llm-emissions-tracker-db
npx nx dev llm-emissions-tracker

# Stop the database
npx nx down llm-emissions-tracker-db

# Run all tests
npx nx run-many -t test

Software Architecture

The LLM Emissions Tracker project has the following software architecture.

Projects

• llm-emissions-tracker - Java/Quarkus API
• llm-emissions-tracker-db - PostgreSQL database
• llm-emissions-tracker-sdk-python - Python SDK
• llm-emissions-tracker-sdk-javascript - JavaScript SDK

Figure: Software architecture components and their interactions.

This is how users interact with the system.

Figure: Software architecture components and user flows.

LLM Emissions Tracker DB

This is a SQL-based DB which stores all information related to LLM Emissions data. It has the following database schema:

Figure: Database schema.

LLM Emissions Tracker API

This is an API built with Java/Quarkus. It has REST endpoints for retrieving data and a WebSocket connection for streaming Emissions data to the database.

LLM Emissions Tracker SDK

This is a python SDK which acts as a wrapper around the LLM Emissions Tracker API. This makes it easier to send and retrieve LLM Emissions data.