Anna H.

Key Skills

Software

Labelbox

Top Subject Matter

Multimodal benchmark evaluation (video-focused)

Data Quality assurance

RLHF training data generation

Top Data Types

Video

Image

Audio

Top Task Types

RLHF

Fine-tuning

Evaluation/Rating

Prompt + Response Writing (SFT)

Transcription

Question Answering

Freelancer Overview

Data Quality Analyst & Benchmark Specialist (AnuVacon, Remote). Brings 8+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose. Education includes Bachelor of Arts, Lipscomb University and TESOL Certification, ITTT. AI-training focus includes data types such as Video and labeling workflows including Evaluation, Rating, and RLHF.

Labeling Experience

Data Quality Analyst & Benchmark Specialist (Anuttacon, Remote)

Don't discloseVideoFine-tuning

Owned end-to-end quality assurance for 16 benchmark projects by designing test cases and answer sets from scratch across multiple content domains. Conducted comparative preference rankings and pass/fail evaluations using established rubrics to generate RLHF training data that informs model fine-tuning pipelines. Coached and calibrated 105 annotators to maintain consistent scoring quality across distributed teams. • Designed benchmark test cases and answer sets across food, media, art, and lifestyle content. • Performed rubric-based preference ranking and pass/fail evaluation for RLHF training data. • Coordinated with quality analysts to identify systemic errors and refine scoring rubrics. • Served as subject matter expert for video tasks, resolving ambiguities and refining SOPs to reduce recurring errors.

2026 - 2026

AI Trainer & Multimodal Annotator (Anuttacon, Remote)

Don't discloseVideoRLHF

Completed 82 multimodal annotation and evaluation projects across video, audio, and image data to improve model performance. Produced RLHF training data via comparative preference rankings and pass/fail evaluations using established rubrics, feeding model fine-tuning pipelines. Led global calibration sessions to align scoring standards and reduce inter-annotator variance across distributed annotator groups. • Annotated multimodal outputs across video, audio, and image domains. • Conducted comparative preference ranking and pass/fail evaluation to create RLHF training signals. • Facilitated calibration sessions for up to 10 members to reduce scoring variance. • Performed adversarial test cases and failure analysis to surface reasoning gaps and calculate failure rates for engineering teams.

2025 - 2026

Education

Work History