Phillip Minkovsky

Key Skills

Software

Data Annotation Tech

Remotasks

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Document

Text

Top Task Types

Classification

Evaluation Rating

Fine Tuning

Freelancer Overview

I work as an AI Prompt Writer at Outlier.ai, where I create and evaluate high-quality prompts across academic and enterprise domains. I develop fine-grained scoring criteria, assess model outputs with precision and clearly differentiate between good and ideal responses. My work consistently improves model performance through detailed analysis, prompt refinement, and rigorous quality control.

Labeling Experience

Fine-Grained Criteria

Data Annotation TechTextEvaluation RatingPrompt Response Writing SFT

In this project, I was presented with a system prompt for a specific enterprise. The user then gave a single or multi-step prompt for which there were multiple model responses. My task was to create fine-grained criteria with which to grade these models, and decide, based on the criteria, which of the models answered the prompts better. If none of the models gave an acceptable answer, I would have to edit the better response to produce an ideal response.

In this project, I was presented with a system prompt for a specific enterprise. The user then gave a single or multi-step prompt for which there were multiple model responses. My task was to create fine-grained criteria with which to grade these models, and decide, based on the criteria, which of the models answered the prompts better. If none of the models gave an acceptable answer, I would have to edit the better response to produce an ideal response.

2024 - 2024

Science & Mathematics Prompt Writing

Scale AITextPrompt Response Writing SFT

In this project, I wrote prompts that stumped the model in its response. In Biology and Mathematics, there tends to be one correct final answer. If the model gave an incorrect final answer, my task was successful, and I would then be tasked with correcting the model in its response.

In this project, I wrote prompts that stumped the model in its response. In Biology and Mathematics, there tends to be one correct final answer. If the model gave an incorrect final answer, my task was successful, and I would then be tasked with correcting the model in its response.

2024

Russian Language Voice Prompts

Scale AIAudioTranslation LocalizationData Collection

In this project my task was to record voice memos in various locations to collect data in Russian.

In this project my task was to record voice memos in various locations to collect data in Russian.

2024

Education

Work History