COLONYAI
Documentation

ColonyAI Developer Docs

API Reference · Architecture · Deployment · ML Pipeline · User Guide

Introduction

ColonyAI — Automated Computer Vision for Microbiology

What is ColonyAI

ColonyAI is a production-grade, automated computer vision platform designed for clinical and industrial microbiology laboratories. It replaces manual colony counting with a YOLOv8-powered detection pipeline, delivering ISO-17025-compliant results with full audit traceability and zero-trust security architecture.

ColonyAI is actively used in accredited laboratory environments. All analysis outputs include uncertainty budgets per ISO 4833-1:2013.

Key Features

FeatureDetails
5-Class Detectioncolony_single, colony_merged, bubble, dust_debris, media_crack
CFU/ml CalculationAuto-computed with dilution factor and plated volume
ISO-17025 ComplianceFull audit trail, uncertainty budget, method validation
SimulatorTransient analysis — test parameters without saving to DB
Role-Based Access5 roles: super_admin, admin, manager, auditor, analyst
PDF/Excel ExportAnnotated plate images, detection tables, CFU summary
i18nEnglish and Bahasa Indonesia UI
Zero-Trust SecurityJWT + refresh tokens, ClamAV file scanning, S3 storage

Tech Stack

LayerTechnology
FrontendNext.js 14 App Router, Tailwind CSS, Lucide Icons, Zustand
BackendFastAPI (async), SQLAlchemy async, Alembic, Pydantic v2
ML EngineYOLOv8n (fine-tuned), Ultralytics, OpenCV, PyTorch
DatabasePostgreSQL 15+ with async driver (asyncpg)
StorageS3-compatible (MinIO / Cloudeka Object Storage)
AuthJWT HS256 access + refresh tokens, bcrypt password hashing
File SecurityClamAV antivirus scanning on every upload
DeploymentDocker Compose, NGINX reverse proxy, Redis cache

Compliance

ColonyAI implements ISO-17025 requirements for automated colony counting methods, including measurement uncertainty, method validation records, and tamper-evident audit logs. Every analysis record is immutable once created — modifications generate new revision entries.