Tirth Patel

Key Skills

Software

Appen

Data Annotation Tech

Remotasks

Toloka

Telus

Scale AI

Top Subject Matter

AI model development and optimization for software engineering (e.g., code generation, bug detection)"

Natural Language Processing (NLP) evaluation and fine-tuning for multilingual developer tools (e.g., coding assistants in multiple languages

Data labeling for code review and programming language model training (e.g., function calling, syntax detection)"

Top Data Types

Audio

Computer Code Programming

Text

Top Task Types

Prompt Response Writing SFT

Question Answering

Segmentation

Text Generation

Translation Localization

Freelancer Overview

I have gained significant experience in data labeling and AI training across several platforms, including Outlier, Remotask, Telus International, Labelbox, and Soul AI. At Outlier and Remotask, I specialized in annotating large datasets for autonomous vehicle systems and natural language processing (NLP) models. My work involved fine-tuning models for tasks like object detection, sentiment analysis, and text classification, ensuring that the AI systems were accurate and responsive to real-world conditions. At Telus International and Soul AI, I contributed to multilingual NLP training, evaluating and improving large language models (LLMs) in various languages, including [your native language]. My role at Labelbox included managing complex image and video annotation projects, collaborating with cross-functional teams to ensure high-quality, efficient labeling processes. This blend of hands-on experience with various tools and industries has sharpened my ability to contribute to diverse AI training projects, setting me apart in this dynamic field.

Labeling Experience

Code Annotation for Automated Code Review System

AppenComputer Code ProgrammingClassificationComputer Programming Coding

This project focused on annotating a large dataset of source code to enhance an automated code review system. The tasks included labeling function calls, detecting and annotating syntax errors, and classifying code snippets into various categories (e.g., sorting algorithms, data manipulation). Over 20,000 code samples were annotated using CVAT and Prodigy, with a focus on accuracy and consistency to improve the AI model’s ability to identify potential issues and provide meaningful feedback. Quality measures included rigorous testing and validation of annotations, as well as regular updates to the labeling guidelines based on feedback from the development team.

This project focused on annotating a large dataset of source code to enhance an automated code review system. The tasks included labeling function calls, detecting and annotating syntax errors, and classifying code snippets into various categories (e.g., sorting algorithms, data manipulation). Over 20,000 code samples were annotated using CVAT and Prodigy, with a focus on accuracy and consistency to improve the AI model’s ability to identify potential issues and provide meaningful feedback. Quality measures included rigorous testing and validation of annotations, as well as regular updates to the labeling guidelines based on feedback from the development team.

2023 - 2023

Multilingual Text Categorization for Customer Support Systems

Scale AITextClassificationText Generation

This project involved categorizing and annotating customer support tickets in multiple languages to enhance a language model's ability to understand and respond to diverse queries. The scope included labeling text for entities (e.g., product names, issue types), classifying tickets based on urgency and topic, and performing sentiment analysis to gauge customer satisfaction. Over 30,000 text samples were processed using Scale AI ,Labelbox and Appen, ensuring accurate labeling across various languages. Quality control measures included periodic reviews, inter-annotator agreement checks, and continuous feedback to maintain high data integrity.

This project involved categorizing and annotating customer support tickets in multiple languages to enhance a language model's ability to understand and respond to diverse queries. The scope included labeling text for entities (e.g., product names, issue types), classifying tickets based on urgency and topic, and performing sentiment analysis to gauge customer satisfaction. Over 30,000 text samples were processed using Scale AI ,Labelbox and Appen, ensuring accurate labeling across various languages. Quality control measures included periodic reviews, inter-annotator agreement checks, and continuous feedback to maintain high data integrity.

2023 - 2023

Education

Work History