Matteo Valentini

Key Skills

Software

Mindrift

OneForma

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Text

Video

Top Task Types

Audio Recording

Classification

Data Collection

Text Generation

Text Summarization

Freelancer Overview

I have extensive experience in data labeling and AI training-data workflows, having worked with Scale AI, OneForma, and Mindrift. At Scale AI, I contributed to large-scale annotation projects involving images, video, audio, and text, abiding by strict quality control protocols and taxonomy standards. I’m skilled with human-in-the-loop pipelines, where automated or pre-labels are refined by human reviewers, and I’m comfortable maintaining consistency, precision, and speed under high data volumes. Working with OneForma sharpened my ability to follow detailed guidelines, perform cross-lingual text/audio transcription & annotation, and navigate certification and onboarding workflows. With Mindrift (Toloka’s partner platform), I helped generate high-quality LLM training data, using domain expertise to ensure accuracy, safe/ethical content, and rigorous review. What sets me apart: strong attention to detail; ability to adapt quickly to different annotation tools and taxonomy requirements; experience doing both micro-tasks and longer form tasks; strong understanding of what makes data useful to downstream ML—noise reduction, balanced labels, avoiding bias. I’ve consistently delivered clean, usable data that supports model performance, not just meets superficial label counts.

Labeling Experience

Scale AI audio recording

Scale AIAudioAudio Recording

Recorded and reviewed high-quality Italian audio data to support speech recognition and text-to-speech training. Tasks included reading scripted prompts with clear diction, varied pacing, and natural intonation, as well as performing spontaneous conversational speech. Ensured accuracy of transcriptions, alignment of timestamps, and overall audio clarity, contributing to the development of robust Italian ASR (automatic speech recognition) and voice synthesis models.

Recorded and reviewed high-quality Italian audio data to support speech recognition and text-to-speech training. Tasks included reading scripted prompts with clear diction, varied pacing, and natural intonation, as well as performing spontaneous conversational speech. Ensured accuracy of transcriptions, alignment of timestamps, and overall audio clarity, contributing to the development of robust Italian ASR (automatic speech recognition) and voice synthesis models.

2025

Scale AI prompt writing

Scale AITextPrompt Response Writing SFT

Developed diverse, high-quality prompts and corresponding answers to train and evaluate large language models. Responsibilities included designing creative and domain-specific queries, writing clear and contextually accurate responses, and ensuring alignment with style, tone, and task guidelines. Applied critical judgment to craft balanced datasets that tested reasoning, creativity, and factual accuracy, while maintaining consistency and linguistic diversity.

Developed diverse, high-quality prompts and corresponding answers to train and evaluate large language models. Responsibilities included designing creative and domain-specific queries, writing clear and contextually accurate responses, and ensuring alignment with style, tone, and task guidelines. Applied critical judgment to craft balanced datasets that tested reasoning, creativity, and factual accuracy, while maintaining consistency and linguistic diversity.

2024

One Forma

OneformaImageObject Detection

Worked on large-scale computer vision datasets by labeling and categorizing images according to detailed taxonomies. Tasks included bounding box annotation, object classification, and quality review to ensure consistency and accuracy across thousands of samples. Collaborated within a distributed annotation workflow, applying strict QA standards to deliver high-quality training data for image recognition models.

Worked on large-scale computer vision datasets by labeling and categorizing images according to detailed taxonomies. Tasks included bounding box annotation, object classification, and quality review to ensure consistency and accuracy across thousands of samples. Collaborated within a distributed annotation workflow, applying strict QA standards to deliver high-quality training data for image recognition models.

2024

Scale AI

Scale AITextClassificationEmotion Recognition

Contributed to large-scale training of Italian Large Language Models by performing high-quality text annotation, evaluation, and prompt-response generation. Tasks included crafting natural Italian prompts, reviewing and correcting model outputs for grammar, fluency, and cultural relevance, and applying detailed rubrics to ensure alignment with project guidelines. Focused on maintaining consistency across thousands of entries while meeting tight deadlines and accuracy benchmarks.

Contributed to large-scale training of Italian Large Language Models by performing high-quality text annotation, evaluation, and prompt-response generation. Tasks included crafting natural Italian prompts, reviewing and correcting model outputs for grammar, fluency, and cultural relevance, and applying detailed rubrics to ensure alignment with project guidelines. Focused on maintaining consistency across thousands of entries while meeting tight deadlines and accuracy benchmarks.

2023

Education

Work History