Tatiana Brodsky

Key Skills

Software

Mindrift

Scale AI

Internal/Proprietary Tooling

Top Subject Matter

prompt-response writing-rewriting in the medical, healthcare field

LLM evaluation in Russian (general) and English and Russian (medical)

medical data, text, images recognition, detection, labeling

LLM evaluation in Russian (general) and English and Russian (bio,medical, nutrition)

Top Data Types

Image

Medical Dicom

Text

Top Task Types

Evaluation Rating

Object Detection

Prompt Response Writing SFT

Text Generation

Translation Localization

Freelancer Overview

I am a medical doctor with comprehensive degrees and over 20 years of work experience. I am also an accomplished medical researcher, writer, and editor, contributing to both scientific and public domains. My involvement with AI in medicine began approximately 10 years ago when our team collaborated with a skilled mathematics team on big clinical data projects. This provided me with extensive experience in clinical data collection, sorting, labeling, classification, analysis, and interpretation. For example, some of our results: Machine learning for predicting 5-year mortality risks… 2022 DOI:10.15829/1728-8800-2022-2908 From 2023 I have experience as an AI tutor, specializing in medicine and healthcare but not limited to these fields. I have worked as a medical expert writer, prompt engineer, prompt-response writer, and evaluator. I also have experience in voice response evaluation, image labeling and evaluation, LLM responses factuality and quality evaluation.

Labeling Experience

Images Quality Evals

Scale AIImageClassificationObject Detection

quality evaluation of two image responses to a prompt with classification, detail detection, description, and comparison

quality evaluation of two image responses to a prompt with classification, detail detection, description, and comparison

2024

Conversation Mode AI Voice Response Rating

Scale AIAudioClassificationTranslation Localization

Comparison of the quality of different variants of LLM voice responses. Evaluation of the model's ability to recognize the voice of the user's request. Evaluation of the adequacy of the response, its factuality, harmlessness, and usefulness.

Comparison of the quality of different variants of LLM voice responses. Evaluation of the model's ability to recognize the voice of the user's request. Evaluation of the adequacy of the response, its factuality, harmlessness, and usefulness.

2024 - 2024

Medical project

MindriftTextText GenerationTranslation Localization

As an AI Tutor Domain Expert in Medicine at Mindrift, I develop specialized content that trains AI models to be ethical, accurate, and domain-specifically responsible. This pivotal work supports the reliability of AI applications in Medicine. Responsibilities include crafting, editing, assessing, and fact-checking AI interactions based on rigorous research.

As an AI Tutor Domain Expert in Medicine at Mindrift, I develop specialized content that trains AI models to be ethical, accurate, and domain-specifically responsible. This pivotal work supports the reliability of AI applications in Medicine. Responsibilities include crafting, editing, assessing, and fact-checking AI interactions based on rigorous research.

2023 - 2024

Machine learning for predicting mortality risks

Internal Proprietary ToolingMedical DicomSegmentationClassification

These were a series of interrelated projects over several years to develop models for assessing mortality risks in human populations with various variations of risk factors, including modifiable and non-modifiable ones. The projects involved a large number of participants with repeated collections of medical data and preparation of databases for machine learning models, then training and validation of the results. The databases included millions of data and measurements, including numeric, text and image. Example of a database: https://new.fips.ru/registers-doc-view/fips_servlet?DB=DB&DocNumber=2020621296&TypeFile=html Labeling included data collection, classification, transcription, segmentation, key point annotation and interpretation. An interesting part of these projects was the standardization of assessment and detailed classification and ranking of data for ML purposes.

These were a series of interrelated projects over several years to develop models for assessing mortality risks in human populations with various variations of risk factors, including modifiable and non-modifiable ones. The projects involved a large number of participants with repeated collections of medical data and preparation of databases for machine learning models, then training and validation of the results. The databases included millions of data and measurements, including numeric, text and image. Example of a database: https://new.fips.ru/registers-doc-view/fips_servlet?DB=DB&DocNumber=2020621296&TypeFile=html Labeling included data collection, classification, transcription, segmentation, key point annotation and interpretation. An interesting part of these projects was the standardization of assessment and detailed classification and ranking of data for ML purposes.

2017 - 2023

Education

Work History