Brian M.

Key Skills

Software

Appen

Data Annotation Tech

Labelbox

Mindrift

OneForma

Remotasks

Scale AI

Toloka

Top Subject Matter

prompt and response writing

image classification

physics

Top Data Types

Computer Code Programming

Image

Text

Top Task Types

Classification

Prompt + Response Writing (SFT)

Question Answering

Relationship

Text Generation

Freelancer Overview

Having worked on many projects in data labeling and AI training data preparation over the years, I am able to develop quality datasets for machine learning algorithms. Some of my skills are making annotations to the data in text, image, and video form, paying attention to the need to meet certain objective of an AI application. This would apply where data labeling was crucial due to project challenges that included natural language processing (NLP), computer vision, as well as autonomous systems. The skills within this role are ability to detail, organizational ability, and ability to maintain high levels of work quality. I know how to use labelbox, super-annotate, and in-house annotation tools that are developed in the firm. Furthermore, I am able to possess a team attitude with AI scientists and programmers to improve the datasets also to enhance the result of training. With years of experience in the agile environment and the ability to meet different project’s demands, I consider myself valuable in any AI program.

Labeling Experience

maths

Scale AITextEvaluation/Rating

During a mathematics-orientated project on Remotasks, I engaged in labeling and verifying datasets relevant to training the AI models in issue-solving, as well as identifying numerical patterns. The activities included typing maths equations, providing mathematical labels, and checking whether the solution procedures are correct and logical. Tags present on a problem level included problem type, such as algebra, calculus, or geometry, symbols and expressions, and data labels to meet mathematical formatting. The project period was three months during which the authors manually annotated the dataset of 20,000 data points containing text and handwritten mathematical expressions. Annotation was done following strict protocols of annotation, peer validation and check by team leaders occasionally hence making it possible to maintain the quality of coding. From this initiative, their model was enhanced and strengthened to solve and analyze mathematical problems using artificial intelligence.

2023

data annotation

Data Annotation TechTextPrompt + Response Writing (SFT)

As there is no external public data set with a large number of annotated AI prompts and responses, a new data set of fifty thousand annotated pairs of Prompt and Response was generated for this project to serve aims of establishing conversational Intelligent Agent for use across sectors such as customer support, virtual assistants, and educational tools. The primary tasks were developing a set of complex, realistic scenarios of a user’s behavior that need to be addressed, creating proper responses based on the context of the conversation, and adding tag such as sentiment, intent, and topic to the data. Extra effort was made to label conversational detail till context like sarcasm, humor, and empathy to enable better understanding of context by Artificial Intelligence. To avoid compromising on quality several checks were employed, which include; Specific operational guidelines, peer review at different levels, and regularly updated and adjusted checklists to reduce any existing biases.

2023

Education