Edu A.

Key Skills

Software

Appen

Labelbox

Data Annotation Tech

Micro1

Scale AI

OneForma

Dataturk

Lionbridge

Toloka

Top Subject Matter

LLM personalization validation

evidence extraction auditing

rubric-based evaluation

Top Data Types

Text

Video

Document

Image

Top Task Types

Red Teaming

RLHF

Freelancer Overview

Oracle Senior Reviewer and Auditor - Outlier AI. Brings 13+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include DataAnnotation, Micro1, and Outlier AI. Education includes Bachelor of Science, Universidad de Chile (2012). AI-training focus includes data types such as Text, Video, and Document and labeling workflows including Evaluation, Rating, and Red Teaming.

Labeling Experience

Linguistic AI Training Expert & Reviewer (Micro1) — Video Diarization & QA + Audio Captioning (2025–Present)

Micro1Video

Executed precision-timestamped dialogue tagging and QA to support client-facing high-fidelity training batches. Produced structured speaker profiling with exhaustive physical and vocal descriptors to handle overlapping audiovisual segments. Authored detailed ambient and specific soundscape captions by translating dense multi-modal information into structured text. • Performed video diarization with precise timestamping • Quality-checked training data for speaker overlap scenarios • Captured audio details beyond standard transcription • Converted multi-modal signals into training-ready text descriptions

2025 - Present

AI Evaluator (DataAnnotation) — LLM Personalization Validation & Evidence/Extraction Analysis (2025–Present)

Text

Conducted rigorous multi-dimensional evaluations to validate LLM personalization architectures. Audited evidence-extraction mechanisms to ensure logical consistency and nuance between model profile statements and underlying sources. Built evaluation rubrics for complex single- and multi-turn user prompt scenarios, including conditional and nested constraints. • Assessed persona alignment across diverse prompts and edge cases • Checked fidelity of contextual adaptation using source evidence • Designed grading criteria for system prompts and image-related tasks • Evaluated instruction handling under multi-step constraint structures

2025 - Present

Oracle Senior Reviewer and Auditor - Outlier AI

AppenImageRed Teaming

You served as a senior reviewer and auditor focused on AI evaluations, rubric architecture, and semantic alignment. You maintained extremely high audit-quality scores while analyzing complex model outputs for subtle logical issues. You applied methods for safety red teaming and personalization validation while translating requirements into gold-standard evaluation rubrics. • Maintained a perfect 5.0/5.0 audit score across 70+ senior tasks requiring deep reasoning and structural analysis. • Engineered localized prompts and designed core rubrics defining gold-standard model training. • Audited model responses to detect logical fallacies and improve reasoning density. • Stress-tested guardrails with adversarial prompting to mitigate jailbreaks and bias while validating user-context personalization.

2024 - Present

Oracle Senior Reviewer & Auditor (Outlier AI) — RLHF & Semantic Alignment + Safety & Personalization (2024–Present)

TextRed Teaming

Refined reward systems for RLHF-style alignment by evaluating truthfulness, instruction following, and safety across complex personas. Conducted adversarial stress tests to identify and mitigate jailbreaking behaviors and bias. Validated that models effectively leverage user-specific context and tool integration for tailored responses. • Provided feedback signals for truthfulness/instruction/safety improvements • Tested guardrails with high-entropy adversarial prompts • Performed bias and jailbreak mitigation evaluation • Assessed personalization capabilities in context-aware response generation

2024 - Present

Oracle Senior Reviewer & Auditor (Outlier AI) — AI Evaluations & Rubric Architecture + RLHF/Semantic Alignment (2024–Present)

Text

Served as an Oracle senior reviewer and auditor for AI evaluations, rubric architecture, and reasoning-quality checks. Maintained consistently high audit performance by reviewing deep reasoning and structural analysis tasks. Engineered localized prompts and core rubrics to define gold-standard model training behaviors. • Audited triple-output responses to detect subtle logical fallacies • Refined localized prompts to standardize evaluation • Designed rubric architecture for gold-standard training signals • Achieved perfect audit scoring across large sets of tasks

2024 - Present

Show more

Education

Work History