Valerio Mariani

Key Skills

Software

Appen

CrowdSource

Labelbox

Remotasks

Scale AI

Other

Top Subject Matter

LLM evaluation in Italian and English

Prompt responses reviewing based on images

LLM evaluation with vocal interaction

Top Data Types

Audio

Image

Text

Top Task Types

Classification

Evaluation Rating

Prompt Response Writing SFT

Question Answering

Text Summarization

Freelancer Overview

• Experienced in diverse LLM training projects (including RLHF, RVFL, safety evaluation, localization), with a strong record of adapting methods to varied objectives and domains. • Academic foundation in Philosophy and Humanities, fostering analytical thinking and interdisciplinary perspectives. • Fluent in English, with full professional proficiency. • Proven track record in cultural organizations and artistic productions, enhancing communication, collaboration, and project delivery. • Freelance writer and columnist, with a focus on cinema and music.

Labeling Experience

Quality Control Internal Auditor

Scale AITextClassificationText Generation

Multiple. (Worked on every Italian project on the platform.)

Multiple. (Worked on every Italian project on the platform.)

2024

AI Trainer

Scale AITextClassificationText Generation

A Rubric project, with evals and RLHF. I worked on both Italian and English rubric projects.

A Rubric project, with evals and RLHF. I worked on both Italian and English rubric projects.

2025 - 2025

AI Trainer

Scale AIImageClassificationText Generation

RLHF on multiturn conversations based on image recognition.

RLHF on multiturn conversations based on image recognition.

2025 - 2025

Team Leader

Scale AITextQuestion AnsweringEvaluation Rating

The scope of the project was to conduct an AB eval on two suggested prompts for a real conversation. There were multiple dimensions to assess. The conversation history was checked as well, to acknowledge the context and identify potential harmful/safety issues. The project lasted many months. Since I was the only "super reviewer" of the team, I managed the entire team supporting the manager.

The scope of the project was to conduct an AB eval on two suggested prompts for a real conversation. There were multiple dimensions to assess. The conversation history was checked as well, to acknowledge the context and identify potential harmful/safety issues. The project lasted many months. Since I was the only "super reviewer" of the team, I managed the entire team supporting the manager.

2024 - 2025

Senior Reviewer

Scale AITextText GenerationRLHF

A SFT RLHF project. Long prompts, long responses. Multiturn. Close attention to details in providing perfect final answers.

A SFT RLHF project. Long prompts, long responses. Multiturn. Close attention to details in providing perfect final answers.

2024 - 2024

Show more

Education

Work History