Fayie M.

Key Skills

Software

Sama

Top Subject Matter

Transport & Logistics-Automous Vehicles

Technlogy -Generative AI

Top Data Types

Text

Video

Image

Top Task Types

Bounding Box

Classification

Text Summarization

Freelancer Overview

I am an Associate AI Data Trainer at Sama with over 4 years of experience specializing in the manual labeling and tagging of diverse datasets including text, images and video used to train and assess machine learning models. My role goes beyond basic labeling. I also review and audit labeled data to ensure accuracy and consistency across datasets. Additionally, I develop, edit and moderate technical and creative content for Generative AI systems, ensuring high quality outputs for Fortune 500 clients. My work includes content moderation, flagging prohibited material such as hate speech and violence, and collaborating with global teams across U.S. time zones to deliver daily reports and real time feedback. What sets me apart is my ability to implement quality control frameworks, use project management tools like Jira and Asana, and work comfortably with CRM systems like Salesforce. I have mentored new content moderators, contributed to refining content moderation policies, and maintained familiarity with data protection laws such as GDPR. My attention to detail, stakeholder collaboration skills, and experience working with technical teams to create top notch datasets make me a strong candidate for advanced data labeling and AI training roles. I am also fluent in English and Kiswahili with basic knowledge of French and Portuguese.

Labeling Experience

Content Moderation

TextClassification

I worked as a content moderator for a leading online employment and company review platform. The scope of the project involved reviewing and moderating user-generated content, including employer reviews, salary submissions, interview experiences, and user comments, to ensure compliance with community guidelines and platform policies. My specific tasks included reviewing submitted content for violations such as fake or biased reviews, hate speech, harassment, off topic posts, promotional material, and personally identifiable information. I flagged and removed content that violated guidelines, escalated complex cases to senior moderators, and documented actions taken for accountability and training purposes. I also responded to user reports and appeals regarding moderated content. The project size was substantial. I reviewed approximately 500 pieces of content per day and 2,500 per week. The platform served millions of users globally, requiring consistent and fair moderation decisions. Quality measures were strictly enforced. I adhered to detailed community guidelines and moderation rubrics. Regular quality audits were conducted where a sample of my decisions was reviewed by quality assurance specialists. I maintained an accuracy target of 98 percent or above, with a low false positive rate (incorrectly removing good content) and false negative rate (missing violations). Feedback sessions were held weekly to address edge cases and guideline updates.

2023 - Present

Data Annotation

VideoBounding Box

I worked on a large scale data annotation project for autonomous vehicle development, focusing on 3D LiDAR data. The scope of the project involved annotating millions of frames captured by LiDAR sensors mounted on test vehicles driving in diverse urban and highway environments. The goal was to train machine learning models to accurately detect, classify and track objects in real time for safe self driving navigation. My specific tasks included drawing tight 3D bounding boxes around objects such as cars, pedestrians, cyclists, trucks, traffic cones, and static infrastructure like lamp posts and barriers. I also assigned classification labels to each object and annotated occlusion and truncation levels to indicate how much of an object was visible. For moving objects, I tracked them across sequential frames to maintain consistent labeling. The project size was substantial, involving thousands of annotated frames per week across multiple geographical regions and weather conditions. Quality measures were strictly enforced. I adhered to annotation guidelines requiring intersection over union (IoU) scores of 70% or higher, meaning bounding boxes had to fit objects with at least 70% accuracy. Regular audits and consensus checks were conducted. My labeled data was randomly sampled and reviewed by senior quality assurance specialists, and any discrepancies triggered feedback loops and retraining sessions. I maintained a weekly quality score target of 90% or above, which I consistently met through careful attention to detail and adherence to project specifications.

2021 - 2023

Education

Work History