Kyle C.

Key Skills

Software

Other

Top Subject Matter

AI Alignment

Natural Language Processing

Computer Vision

Top Data Types

Text

Image

Video

Audio

Document

Top Task Types

RLHF

Classification

Transcription

Freelancer Overview

I am a detail-oriented data annotation and AI training specialist with a background in Information Technology and experience supporting AI workflows, model evaluation, and dataset quality improvement. My work includes evaluating AI-generated responses for coherence, factual accuracy, relevance, and safety, as well as performing image, video, and audio annotation while following strict annotation guidelines. I am skilled in identifying inconsistencies, maintaining labeling accuracy, and delivering structured feedback to improve overall model performance. I work efficiently in remote environments and am committed to producing high-quality annotations with consistency and attention to detail. With strong analytical and communication skills, fluency in English and Tagalog, and experience handling time-sensitive tasks, I am prepared to contribute effectively to data annotation, computer vision, transcription, and AI training projects.

Labeling Experience

AI Training & Annotation Contributor

OtherAudioTranscription

In my role, I transcribed audio data and performed quality review to ensure linguistic and transcriptional accuracy. This work involved converting speech to text for use in conversational AI and speech recognition models. I strictly followed transcription conventions and project requirements throughout the process. • Accurately transcribed varied audio recordings into text format • Performed quality review for transcription accuracy and completeness • Ensured linguistic correctness and adherence to guidelines • Provided datasets supporting speech-to-text and AI model training

Present

AI Training & Annotation Contributor

OtherVideoClassification

I completed video annotation tasks including classification and object labeling on motion-based content. This involved identifying, segmenting, and labeling objects or events within video sequences. My contributions ensured data quality for training and evaluating machine learning models. • Annotated video frames for object detection and classification tasks • Ensured temporal consistency throughout the sequence annotation • Complied with annotation quality standards set by the project • Enabled robust video model training by providing accurate labels

Present

AI Training & Annotation Contributor

OtherImageClassification

I performed image annotation tasks involving both classification and object labeling. As part of my responsibilities, I adhered to strict guidelines to maintain annotation quality and consistency. My work supported the improvement of image recognition and computer vision model performance. • Used established protocols for image classification and annotation • Labeled visual datasets to identify specific objects and features • Maintained accuracy in annotation across all assigned images • Supported the training of computer vision models through high-quality data

Present

AI Training & Annotation Contributor

OtherTextRLHF

As an AI Training & Annotation Contributor for Outlier AI, I evaluated and ranked large language model outputs through RLHF workflows. I reviewed AI-generated responses for relevance, accuracy, coherence, and alignment with safety guidelines. I also refined prompts, conducted detailed annotation, and ensured high data quality standards. • Conducted RLHF-based response evaluation and ranking of text outputs • Identified hallucinations and policy violations in LLM responses • Applied detailed annotation guidelines and maintained consistency • Refined prompts to improve generation quality and controlled output

Present

Aether

ImageRLHF

A dataset consisting of images are provided, the task is to mask an object in the image to remove and then engineer a prompt to add that object back in. You would then do that again for as much objects in the photo as you can, when there is no more workable objects in the image, you would then engineer a prompt that would add all of those back in. There were strict guidelines ie. making sure the image produced does not have artifacts and that the prompts are precise and detailed.

2026 - 2026

Education

Work History