A developer platform that tracks resource usage and CO₂ impact of software features in production. Combines feature flag architecture, telemetry ingestion, and sustainability analytics to give engineering teams visibility into the environmental cost of their code.
the problem
Engineering teams have no visibility into the environmental cost of their software features - CPU cycles, memory, and energy consumption are tracked for performance but never mapped to CO₂ output.
the impact
Carbonly gives developers a drop-in SDK to measure execution time and memory per feature, then converts that telemetry into energy and CO₂ estimates — making sustainability a first-class engineering metric.
technologies used
Language
Backend Framework
Build Tool
Other
Database
Styling
challenges faced
Designing a deterministic, explainable sustainability calculation engine without AI or ML.
Building a high-throughput telemetry ingestion pipeline that handles batched SDK events at scale.
Implementing a multi-tenant architecture with project/environment/API-key isolation similar to feature flag platforms.
solutions implemented
Built a three-step formula engine (compute units -> energy kWh -> CO₂ grams) using industry-standard coefficients from SPEC Power and IEA data.
Implemented a BullMQ worker pipeline for hourly and daily metric rollups with idempotent bucket inserts.
Adopted an Unleash-inspired SDK key model with SHA-256 hashed keys and environment-scoped access.
key features
Drop-in Node.js SDK (@carbonly/node) with Express middleware and NestJS module support.
Real-time sustainability dashboard showing CO₂ grams, energy kWh, and compute units per feature.
Project and environment management with scoped API keys for dev, staging, and production.
BullMQ-powered aggregation pipeline for hourly and daily metric rollups.
outcomes & results
Published @carbonly/node to npm with full TypeScript support, Express middleware, and NestJS integration.
End-to-end pipeline from SDK event -> ingestion -> aggregation -> CO₂ calculation running in production.
key learnings
The complexity of translating raw compute metrics into meaningful sustainability signals for non-specialist audiences.
Designing SDK ergonomics that are zero-config for simple use cases but fully configurable for advanced integrations.
The architectural parallels between feature flag platforms and telemetry ingestion systems.
similar projects
Learning Management System (LMS) for Nursing Students
A learning management system (LMS) designed to provide nursing students with a comprehensive platform for preparing for their licensure exams, offering a wide range of resources and tools to enhance their learning experience.
explore more projects
eFama App - Agricultural Marketplace
A mobile application designed to connect South African farmers and buyers in a seamless marketplace, facilitating the trades of agriciultural products and services.
Learning Management System (LMS) for Nursing Students
A learning management system (LMS) designed to provide nursing students with a comprehensive platform for preparing for their licensure exams, offering a wide range of resources and tools to enhance their learning experience.