Precious K.

Key Skills

Software

Other

Google Cloud Vertex AI

Telus

Mercor

Toloka

Appen

Lionbridge

Clickworker

Top Subject Matter

Video captioning and OCR for AI training datasets

LLM evaluation and quality assessment for AI training

Video quality evaluation and dataset collection for AI training

Top Data Types

Text

Audio

Document

Video

Image

Top Task Types

Object Detection

Text Generation

Question Answering

Computer Programming/Coding

Data Collection

Transcription

Classification

Audio Recording

Freelancer Overview

Mercor — Video Captioning / Optical Character Recognition Expert. Core strengths include Mercor, Toloka, and Appen. Education includes Bachelor of Engineering, Michael Okpara University of Agriculture, Umudike (2023) and C1 Advanced English Certificate, EF SET (2025). AI-training focus includes data types such as Video, Text, and Audio and labeling workflows including Transcription, Evaluation, and Rating.

Labeling Experience

Mercor — Generalist Expert

MercorText

Designed and developed prompts to evaluate and stress-test advanced AI model capabilities across varied domains. Reviewed and scored model-generated responses using structured rubrics to provide consistent, reasoned quality assessments. Identified recurring weaknesses in outputs and flagged data inconsistencies to improve iterative model and dataset refinement. • Wrote 50+ prompts for varied-domain model capability testing • Rated 100+ responses using structured rubric criteria • Flagged ambiguities and annotation errors affecting dataset integrity • Provided actionable feedback for iterative improvements

2026 - Present

Mercor — Video Captioning / Optical Character Recognition Expert

MercorVideoTranscription

Transcribed on-screen text from short-form video content while strictly following formatting requirements and annotation guidelines. Assigned accurate start and end timestamps to each extracted text element to ensure structured alignment for downstream AI model training. Maintained consistent quality across high volumes of videos to support reliable dataset creation. • Transcribed visible text content with attention to punctuation and formatting • Timestamped each text element using provided temporal rules • Followed detailed project annotation guidelines for clean structured output • Ensured dataset readiness for downstream processing and model use

2026 - Present

Mindrift (Toloka) — Video Annotation & Data Collection

TolokaVideo

Conducted comparative evaluations of videos to assess quality, distortion level, and adherence to original prompts. Reviewed side-by-side clips to determine which segments best matched prompt requirements and provided structured feedback for model training. Collected first-person household task workflow video datasets by following strict project specifications for camera orientation, hand visibility, and task continuity. • Evaluated prompt adherence and visual quality across paired videos • Documented which segments best satisfied prompt requirements • Recorded first-person workflow videos with required filming constraints • Maintained dataset integrity by adhering to recording standards

2025 - Present

Clickworker — UHRS Evaluator

ClickworkerText

Completed microtasks involving collection and labeling of data across text, audio, image, and video formats for machine learning training datasets. Rated search engine results for relevance, quality, and helpfulness based on defined evaluation criteria. Performed classification and tagging and executed tasks including image uploads and voice recording workflows. • Collected, cleaned, and labeled multimodal data for ML training • Rated search results using relevance and helpfulness criteria • Performed image upload and item classification/tagging tasks • Completed voice recording and related evaluation microtasks

2022 - Present

IoT & Embedded Systems Instructor (Deep STEM)

AppenVideoClassification

Designed and delivered STEM engineering instruction that translates complex topics into clear explanations for learners at multiple proficiency levels. While not explicitly framed as an AI-annotation job, this work closely maps to producing and calibrating high-quality technical responses used in AI training and evaluation. Focus areas included guiding students through embedded/IoT concepts and developing structured learning materials suitable for prompt/response quality checks. • Developed course materials and step-by-step project guides for home automation, robotics, and sensor integration • Taught ESP8266/Arduino/microcontroller logic in plain language to non-experts • Built 10+ hands-on exhibition projects with detailed technical documentation • Calibrated explanation depth based on learner background and comprehension needs

2024 - 2025

Show more

Education

Work History