Benard K.

Key Skills

Software

Surge AI

Toloka

Top Subject Matter

LLM / Generative AI response evaluation and RLHF support

Audio dataset annotation

Transcription Domain Expertise

Top Data Types

Text

Audio

Image

Document

Top Task Types

Transcription

Freelancer Overview

Gen AI Model Trainer and AI Evaluation Specialist — Surge AI (Remote). Core strengths include Surge AI and Toloka. Education includes Bachelor of Science, University of Central Arkansas. AI-training focus includes data types such as Text and Audio and labeling workflows including Evaluation, Rating, and Transcription.

Labeling Experience

Gen AI Model Trainer and AI Evaluation Specialist — Surge AI (Remote)

Surge AIText

Evaluated and ranked LLM responses for helpfulness, factuality, safety, instruction-following, reasoning quality, and tone to support model alignment. Performed comparative preference ranking and identified hallucinations, logical inconsistencies, edge-case failures, and bias risks using nuanced evaluation rubrics. Produced actionable evaluator feedback to guide future training and optimization priorities for reasoning- and safety-sensitive outputs. • Ranked and scored model responses against multi-dimensional rubrics • Conducted preference ranking and comparative response analysis for RLHF workflows • Documented failure patterns and bias concerns for downstream model improvement • Assessed adherence to multi-step instructions and safety constraints

2025 - 2026

Data annotator — Alignerr AI (Remote)

TolokaAudioTranscription

Annotated and reviewed large-scale audio datasets for AI training, including ASR transcription, speaker diarization, and intent classification. Applied computer vision labeling tools and techniques (bounding boxes, polygons, segmentation, and key-point labeling) to validate complex multimodal datasets. Performed dataset validation to find duplicates, labeling errors, and taxonomy inconsistencies, improving data quality and consistency across gold-standard evaluations. • Completed ASR transcription, speaker tagging/diarization, and intent classification annotations • Produced computer vision annotations including bounding boxes, polygons, segmentation, and key-point labels • Ran calibration tasks, quality audits, and dataset validation for duplicates and taxonomy errors • Documented edge cases and refined guideline interpretations to improve evaluator agreement

2025 - 2025

AI Data Annotator and Content Evaluator — Toloka (Remote)

TolokaText

Performed AI training annotation and content evaluation for text-based tasks including response ranking, quality review, and guideline-based benchmarking. Assisted with content moderation and policy compliance by applying safety classification and verifying annotation consistency across datasets. Supported multilingual and localization-focused evaluation workflows while collaborating with distributed teams to maintain accuracy and turnaround performance. • Evaluated AI-generated outputs using detailed project guidelines and benchmarks • Conducted editorial QA to ensure consistency of labels and annotations • Applied safety/policy classification for content moderation and compliance • Supported multilingual/localization AI evaluation and dataset validation tasks

2024 - 2025

Education

Work History