Bruno Azevedo

Key Skills

Software

Label Studio

Scale AI

Surge AI

Top Subject Matter

Software Engineering

PR Review Automation

Natural Language Processing

Top Data Types

Text

Document

Top Task Types

Text Summarization

Data Collection

Freelancer Overview

AI Trainer – PR-Writer Project (Revelo). Brings 11+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include Label Studio, Scale AI, and Surge AI. Education includes Bachelor of Science, Federal University of Rio de Janeiro (2016) and Non-Degree Studies, Kanazawa University (2014). AI-training focus includes data types such as Text, Computer Code, and Programming and labeling workflows including Text Summarization, Computer Programming, and Coding.

Labeling Experience

AI Trainer – Behavioral Data Architecture Project (Revelo)

Label StudioTextData Collection

Designed scalable behavioral data collection frameworks and managed annotation pipelines for multi-turn conversational AI training. Structured data architecture decisions around schema, taxonomy, and storage for large-scale labeled datasets. Aligned outputs with model evaluation benchmarks and acceptance criteria. • Implemented frameworks for annotating complex conversational data. • Developed schemas and label taxonomy for optimal data storage. • Coordinated pipeline creation with engineering and ML teams. • Ensured labeled datasets met model training and safety goals.

Designed scalable behavioral data collection frameworks and managed annotation pipelines for multi-turn conversational AI training. Structured data architecture decisions around schema, taxonomy, and storage for large-scale labeled datasets. Aligned outputs with model evaluation benchmarks and acceptance criteria. • Implemented frameworks for annotating complex conversational data. • Developed schemas and label taxonomy for optimal data storage. • Coordinated pipeline creation with engineering and ML teams. • Ensured labeled datasets met model training and safety goals.

2024 - 2026

AI Trainer – Behavioral Data Debugging Project (Revelo)

Label StudioText

Analyzed and debugged model behavioral data to identify unsafe outputs, inconsistencies, and annotation errors in LLM pipelines. Ranked model responses on helpfulness, harmlessness, and honesty for RLHF datasets. Enabled targeted retraining and rubric refinement through systematic error tracing. • Conducted behavioral data audits and error pattern detection. • Ranked LLM outputs for RLHF training cycles. • Collaborated in refining training and evaluation rulesets. • Supported data correction loops for enhanced model performance.

Analyzed and debugged model behavioral data to identify unsafe outputs, inconsistencies, and annotation errors in LLM pipelines. Ranked model responses on helpfulness, harmlessness, and honesty for RLHF datasets. Enabled targeted retraining and rubric refinement through systematic error tracing. • Conducted behavioral data audits and error pattern detection. • Ranked LLM outputs for RLHF training cycles. • Collaborated in refining training and evaluation rulesets. • Supported data correction loops for enhanced model performance.

2024 - 2026

AI Trainer – IaC Annotation Project (Revelo)

Label Studio

Annotated Infrastructure-as-Code scripts for semantic correctness, security, and best-practice compliance to support AI model training. Defined labeling schemas for various IaC resource types and configuration intents with domain experts. Detected and escalated code snippet misclassifications to improve annotated dataset integrity. • Labeled scripts in Terraform, Ansible, and CloudFormation. • Specified dependency mappings and configuration intent in code datasets. • Conducted code-level quality assurance and schema validations. • Communicated with subject matter experts to develop precise labeling protocols.

Annotated Infrastructure-as-Code scripts for semantic correctness, security, and best-practice compliance to support AI model training. Defined labeling schemas for various IaC resource types and configuration intents with domain experts. Detected and escalated code snippet misclassifications to improve annotated dataset integrity. • Labeled scripts in Terraform, Ansible, and CloudFormation. • Specified dependency mappings and configuration intent in code datasets. • Conducted code-level quality assurance and schema validations. • Communicated with subject matter experts to develop precise labeling protocols.

2024 - 2026

AI Trainer – PR-Writer Project (Revelo)

Label StudioTextText Summarization

Labeled and curated thousands of pull request descriptions to train NLP models for auto-generating PR summaries and classifying code changes. Developed annotation guidelines, validated model outputs, and reduced disagreement by standardizing decision frameworks. Contributed to both manual and automated quality assurance workflows for technical document labeling. • Curated training sets for PR summarization and intent classification. • Established and documented annotation protocols for consistent labeling. • Validated AI-generated outputs for hallucinations and factual correctness. • Collaborated with teams to ensure high-quality data for LLM fine-tuning.

Labeled and curated thousands of pull request descriptions to train NLP models for auto-generating PR summaries and classifying code changes. Developed annotation guidelines, validated model outputs, and reduced disagreement by standardizing decision frameworks. Contributed to both manual and automated quality assurance workflows for technical document labeling. • Curated training sets for PR summarization and intent classification. • Established and documented annotation protocols for consistent labeling. • Validated AI-generated outputs for hallucinations and factual correctness. • Collaborated with teams to ensure high-quality data for LLM fine-tuning.

2024 - 2026

Education

Work History