Robera Belachew

Key Skills

Software

HiveMind

Top Subject Matter

No subject matter listed

Top Data Types

Image

Top Label Types

Classification

Question Answering

Object Detection

Action Recognition

Mapping

Freelancer Overview

I am an experienced AI operations specialist and software engineer with a strong background in data labeling, annotation, and AI training data workflows. My expertise includes auditing large language model outputs, semantic scoring, and mitigating hallucinations to ensure high-quality, reliable datasets for NLP applications. I specialize in Oromo ↔ English translation and cultural localization, contributing native-level linguistic insights to multilingual AI projects. I have designed and optimized human-in-the-loop validation processes and built custom automations using tools like Zapier, Make, and Airtable to streamline annotation pipelines. My technical skills in SQL, Python, and Excel enable me to synthesize complex data and drive actionable improvements. Through projects like financial compliance automation and CRM workflow optimization, I have demonstrated a commitment to advancing ethical, high-precision AI systems.

Labeling Experience

HiveAI

HivemindImageClassificationQuestion Answering

Upload Image + JSON File (Desktop Only) Desktop-only data submission task: Upload screenshots (images) paired with their corresponding structured JSON files from your computer. This supports the creation of multimodal datasets for vision-language AI models. Each task requires accurate file matching, clean captures without personal information, and verification. Designed for contributors with a reliable desktop or laptop setup and stable browser environment. High task volume expected with long-term availability. Type: Image and File Upload Requirements: Desktop computer only; complete qualification test. Controversial Content Categorization Content safety and moderation task: Review images together with their associated text prompts, then classify whether the pair is controversial, harmful, explicit, or neutral/safe. Follow the provided guidelines and cheatsheet for consistent evaluation of sensitive topics such as violence, adult content, or misinformation. Contributes to training AI m

Upload Image + JSON File (Desktop Only) Desktop-only data submission task: Upload screenshots (images) paired with their corresponding structured JSON files from your computer. This supports the creation of multimodal datasets for vision-language AI models. Each task requires accurate file matching, clean captures without personal information, and verification. Designed for contributors with a reliable desktop or laptop setup and stable browser environment. High task volume expected with long-term availability. Type: Image and File Upload Requirements: Desktop computer only; complete qualification test. Controversial Content Categorization Content safety and moderation task: Review images together with their associated text prompts, then classify whether the pair is controversial, harmful, explicit, or neutral/safe. Follow the provided guidelines and cheatsheet for consistent evaluation of sensitive topics such as violence, adult content, or misinformation. Contributes to training AI m

2023 - 2025

Education

Work History