Christabel S.

Key Skills

Software

Micro1

Toloka

Telus

Top Subject Matter

Egocentric video datasets for computer vision and robotics

LLM response evaluation and rubric-based benchmarking

RLHF evaluation and multimodal dataset contribution for computer vision and robotics

Top Data Types

Video

Text

Image

Top Task Types

Segmentation

RLHF

Data Collection

Classification

Freelancer Overview

Luel — Review and correction of segment-level egocentric video annotations. Brings 6+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Annotation platforms, Google Sheets, and Google Docs. Education includes Bachelor of Science, Kwame Nkrumah University of Science and Technology (2023). AI-training focus includes data types such as Video and Text and labeling workflows including Segmentation, Evaluation, and Rating.

Labeling Experience

Toloka Annotators — LLM response evaluation and rubric grading

Text

Evaluated large language model responses by applying predefined guidelines and rubrics to compare and assess output quality. Synthesized information across large volumes of model responses, flagging edge cases and recurring failure modes. Produced structured written rationales to support benchmarking and inform product and model refinement. • Compared LLM outputs against rubric-based guidelines. • Flagged edge cases and identified recurring patterns or failure modes. • Provided detailed structured justification for scoring decisions. • Performed quality-focused benchmarking support and QA documentation.

2026 - Present

Luel — Review and correction of segment-level egocentric video annotations

VideoSegmentation

Reviewed and corrected segment-level annotations for egocentric video datasets while validating timestamp alignment and labeled event sequences. Ensured consistency between textual labels and observed visual actions, and identified annotation errors against quality standards. Enforced reliable, coherent annotations across labeled events to support downstream computer vision and robotics training. • Verified timestamps, action sequences, and object interactions within labeled events. • Checked agreement between textual descriptions and corresponding visual actions. • Detected labeling inaccuracies and applied annotation quality guidelines. • Maintained annotation consistency for segment-level video labels.

2026 - Present

Micro1 — RLHF-style evaluation and multimodal training data annotation

VideoRLHF

Performed rubric-based grading and RLHF-style evaluation to improve model response quality. Contributed to multimodal training datasets used for computer vision and robotics workflows, including structured recording of real-world task sequences. Applied iterative feedback loops to increase annotation quality and consistency across collected data. • Conducted RLHF-style evaluation and rubric-based grading of responses. • Contributed multimodal training datasets for vision and robotics. • Recorded structured task sequences emphasizing human-object interactions. • Used iterative feedback to refine annotation consistency and quality.

2026 - 2026

Pedasoft Consult — clustering/segmentation analytics supporting model-driven decisions

VideoData Collection

Developed AI-related data preparation and analytics workflows that supported model development rather than direct labeling production. Implemented clustering and dimensionality reduction to derive structured customer segments, improving downstream decision-making. Although not explicitly described as labeling, the work produced structured training/analysis datasets via data preprocessing and transformation. • Built customer segmentation using clustering algorithms. • Applied PCA for dimensionality reduction to extract key features. • Enhanced targeted strategies by deriving distinct customer segments. • Visualized and interpreted results for actionable insights.

2025 - 2026

BrantSMS — AI model evaluation and quality assurance for response generation

Text

Evaluated AI model responses to user queries across diverse computer science topics, applying quality checks to maintain relevance. Conducted data validation and quality assurance to ensure model integrity and improve response quality. Collaborated with cross-functional teams to refine AI-generated content based on evaluation outcomes. • Assessed AI response quality for queries across computer science topics. • Performed data validation and QA for model integrity. • Worked with teams to improve response relevance. • Supported maintenance of large-scale SQL data for model workflows.

2021 - 2024

Education

Work History