Sonny Boy Jr Aniceto

Key Skills

Software

Labelbox

Remotasks

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Document

Text

Top Task Types

Evaluation Rating

Prompt Response Writing SFT

Question Answering

RLHF

Freelancer Overview

I have extensive experience in training AI models, particularly in the domains of physics and mathematics. I specialize in guiding AI models to write accurate solutions and arrive at correct answers. This involves designing and engineering training problems at undergraduate and graduate levels, covering advanced topics such as classical mechanics, electromagnetism, nuclear physics, and multivariable calculus. My work ensures that models not only solve problems correctly but also develop a deeper understanding of the underlying principles.

Labeling Experience

Math Reasoning Evaluator

Scale AITextEvaluation RatingPrompt Response Writing SFT

The project involves designing original, challenging problems in university-level mathematics, including calculus, algebra, and statistics, as well as competitive mathematics, aimed at testing and extending the AI model's problem-solving capabilities. As a data labeler, my responsibilities included 1) evaluating the accuracy and validity of the model’s solutions by thoroughly analyzing its reasoning and calculations and 2) revising and improving the model’s solutions to ensure correctness and clarity. This process not only identified gaps in the model's understanding but also contributed to refining its ability to tackle diverse mathematical problems effectively.

The project involves designing original, challenging problems in university-level mathematics, including calculus, algebra, and statistics, as well as competitive mathematics, aimed at testing and extending the AI model's problem-solving capabilities. As a data labeler, my responsibilities included 1) evaluating the accuracy and validity of the model’s solutions by thoroughly analyzing its reasoning and calculations and 2) revising and improving the model’s solutions to ensure correctness and clarity. This process not only identified gaps in the model's understanding but also contributed to refining its ability to tackle diverse mathematical problems effectively.

2023

Physics Reasoning Evaluator

Scale AITextEvaluation RatingPrompt Response Writing SFT

The project involves creating original, challenging problems in pure and applied physics designed to test the limits of the AI model’s problem-solving capabilities. As a data labeler, my role was to 1) evaluate the correctness of the model's solutions by thoroughly analyzing its reasoning and calculations and 2) develop and implement precise revisions to enhance the model’s understanding and performance. This approach ensured not only accurate solutions but also a refined problem-solving methodology within the model.

The project involves creating original, challenging problems in pure and applied physics designed to test the limits of the AI model’s problem-solving capabilities. As a data labeler, my role was to 1) evaluate the correctness of the model's solutions by thoroughly analyzing its reasoning and calculations and 2) develop and implement precise revisions to enhance the model’s understanding and performance. This approach ensured not only accurate solutions but also a refined problem-solving methodology within the model.

2023

Python Code Optimization

LabelboxComputer Code ProgrammingFine TuningEvaluation Rating

The project focused on enhancing the AI model’s efficiency in generating Python code responses. Given a prompt consisting of a Python script and a textual description of the required changes, along with the AI's response in the form of revised code and explanatory text, my role was to evaluate and improve the code component of the AI’s response. Specifically, I identified areas of the response that could be optimized by avoiding redundant or unnecessarily verbose code ("lazy coding") while maintaining functionality and clarity. This process streamlined the AI’s output, reducing response generation time and improving overall efficiency in coding tasks.

The project focused on enhancing the AI model’s efficiency in generating Python code responses. Given a prompt consisting of a Python script and a textual description of the required changes, along with the AI's response in the form of revised code and explanatory text, my role was to evaluate and improve the code component of the AI’s response. Specifically, I identified areas of the response that could be optimized by avoiding redundant or unnecessarily verbose code ("lazy coding") while maintaining functionality and clarity. This process streamlined the AI’s output, reducing response generation time and improving overall efficiency in coding tasks.

2024 - 2024

Education

Work History