Daniele Sorabella

Key Skills

Software

CVAT

Label Studio

Prodigy

Top Subject Matter

No subject matter listed

Top Data Types

Document

Medical Dicom

Text

Top Task Types

Diagnosis

Question Answering

Text Generation

Freelancer Overview

As a board-certified medical doctor and clinical nutrition specialist fluent in both English and Italian, I bring domain-level expertise to AI training projects involving medical, legal, and general language tasks. I have hands-on experience in data labeling, LLM evaluation (RLHF), and multilingual content moderation across platforms such as Toloka, Outlier, and DataAnnotation.tech. My work includes ranking AI-generated outputs, evaluating medical summaries, and classifying sensitive text data with high consistency and accuracy. With over 5 years of experience in clinical research, academic writing, and sworn legal translation, I specialize in delivering precise, guideline-driven annotations across complex subjects. I am confident working under strict quality protocols (95%+ accuracy), and I am comfortable contributing to projects involving healthcare datasets, policy alignment, and multilingual NLP.

Labeling Experience

Medical and Legal Text Evaluation for RLHF Systems

Label StudioTextClassificationData Collection

I contributed to several large-scale AI training projects aimed at improving LLM alignment and safety, particularly in the medical and legal domains. Tasks included: ranking AI-generated responses (RLHF), evaluating text accuracy, identifying bias/harmful outputs, summarizing complex clinical information, verifying translations, and classifying multilingual medical/legal content. I worked in both English and Italian, using detailed task-specific guidelines to maintain consistency and a >95% quality rating across over 1,200+ reviewed items. I also participated in prompt-response refinement and red-teaming activities to detect unsafe or hallucinated medical claims. Projects complied with HIPAA/GDPR standards and required strict adherence to confidentiality and accuracy metrics.

I contributed to several large-scale AI training projects aimed at improving LLM alignment and safety, particularly in the medical and legal domains. Tasks included: ranking AI-generated responses (RLHF), evaluating text accuracy, identifying bias/harmful outputs, summarizing complex clinical information, verifying translations, and classifying multilingual medical/legal content. I worked in both English and Italian, using detailed task-specific guidelines to maintain consistency and a >95% quality rating across over 1,200+ reviewed items. I also participated in prompt-response refinement and red-teaming activities to detect unsafe or hallucinated medical claims. Projects complied with HIPAA/GDPR standards and required strict adherence to confidentiality and accuracy metrics.

2024 - 2024

Education

Work History