Designed for Teams That Require Reliable Annotation.
Aeterna Data is a specialized image annotation company that works with computer vision datasets and visual AI evaluation. We are a small team with structured workflows, and we check the quality of every batch to keep our standards high.
The Problem
The Problem We Solve
Most annotation vendors operate as large crowd-labor platforms, distributing tasks to anonymous workers with minimal quality oversight. This approach can be effective for simple, high-volume labeling. However, for computer vision training datasets that require consistent boundary rules, class-specific guidelines, and measurable quality, it often fails to deliver the necessary standards.
Aeterna Data was created to meet this need. Our small, dedicated team follows clear annotation workflows. We measure IAA for each batch, run a pilot phase to check our process, and set up formal agreements before sharing data. We use the same team and standards for every project.
Comparison
How We Are Different
Crowd Platforms
- ✗Anonymous worker pools
- ✗No dedicated team is assigned to each project
- ✗Quality varies between batches
- ✗No standard exists for formal legal agreements
- ✗Processes are hard to audit or reproduce
Aeterna Data
- ✓Dedicated team for each project
- ✓IAA measured for every batch
- ✓Structured pilot phase before production
- ✓NDA and DPA before data sharing
- ✓Consistent annotators used throughout
Large Enterprise Vendors
- ✗High minimums required
- ✗Long onboarding timelines
- ✗Designed for volume over precision
- ✗Hard to communicate directly
- ✗Not suited to early-stage teams
The Founder
Precision as a Starting Point, Not a Goal.
Muhammad Rifqi Fauzan Arifin
Founder & Director
PT Aeterna Data Intentio Logic
Muhammad Rifqi Fauzan Arifin is the founder and lead annotator of Aeterna Data. With over two years of hands-on experience in annotation pipelines, including team management, bounding box labeling, and RLHF evaluation, he established Aeterna Data on the principle that annotation quality must be measurable rather than assumed.
Before founding Aeterna Data, he expanded an annotation team from two to fifty members in one year at his previous company, where he managed large-scale computer vision labeling and quality validation for retail AI training datasets. His leadership and expertise in annotation now strengthen Aeterna Data's operations.
Quality Standard
Measured. Reported. Guaranteed.
All Aeterna Data annotation batches are held to the same quality standard, regardless of service type or project size. We measure and report inter-annotator agreement as a required threshold with every delivery.
Cohen's Kappa
Pairwise IAA. Applied when two annotators label the same item. Default threshold across all standard production batches.
Fleiss' Kappa
Multi-annotator IAA. Applied when three or more annotators label the same item. Used on complex visual RLHF preference ranking and high-ambiguity annotation tasks.
Pilot Batch First
100–500 records annotated before production begins. Written client approval required before any production work commences.
IAA Every Batch
Cohen's Kappa and Fleiss' Kappa measured on every production batch without exception.
Rework at Zero Cost
Any batch failing either threshold is reworked before delivery. Rework hours are never billed.
Contract Stack
Formal Agreements. Every Project.
Our training datasets are proprietary. At Aeterna Data, we handle every project under a clear legal framework. We always sign an NDA and a DPA before sharing any data.
Non-Disclosure Agreement (NDA)
Mutual. Signed first. Every annotator on your project also signs an individual NDA before accessing your platform.
Data Processing Agreement (DPA)
GDPR Article 28 compliant. Defines how we process, store, and protect your data throughout the engagement.
Data Security Policy
Our internal security standards for handling client datasets. Covers access controls, storage protocols, and annotator device requirements.
Ideal Clients
Who We Work With
Good fit:
- ✓AI startups building computer vision models that need reliable annotation but prefer not to work with large vendors
- ✓Research labs preparing image datasets for model training or benchmarking
- ✓ML teams at growing companies with annotation needs that are too specific for standard crowd platforms
- ✓Teams that have struggled with inconsistent quality from crowd annotation services
Not the right fit:
- ✗Teams seeking to label millions of images and require rapid delivery
- ✗Projects without strict quality standards or detailed annotation guidelines
- ✗Teams that need a full platform, not just an annotation team
Ready to Prove the Pipeline?
Start with a paid pilot. One week. Your dashboard. A complete IAA quality report on delivery. No long-term commitment required. Pilot fee credited to your first invoice.