Atul P.

Key Skills

Software

Other

CrowdSource

Top Subject Matter

No subject matter listed

Top Data Types

Image

Text

Video

Top Task Types

RLHF

Freelancer Overview

I have three months of hands-on experience in Reinforcement Learning from Human Feedback (RLHF), working with Soul AI (Deccan AI) on a Gemini Project. During this time, I have clocked more than 200 hours training the Gemini model across different capacities, including Contextual Understanding, Messaging Tasks, Extensions, and more. Through this work, I developed strong skills in data labeling, output evaluation, and fine-tuning AI responses based on detailed rubrics and project guidelines. What sets me apart is my keen eye for accuracy, relevance, and user intent alignment. I have built a strong foundation in identifying subtle improvements in model behavior and maintaining consistency across diverse tasks. I am confident in my ability to deliver high-quality training data and am eager to continue building my skills in AI development and evaluation.

Labeling Experience

Outlier

OtherAudioFine-tuning

fine tune voice, by adding nuances and help genetrate better responses in hindi

2024

Soul AI

CrowdSourceTextAction Recognition

Label actions performed by AI for google home devices

2024

Education

Work History