For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
D

Dinesh Kanna

LLM Trainer / Delivery Data Analyst (Data Labeling & RLHF Evaluation)

INDIA flag
Madurai, India
$25.00/hrIntermediateInternal Proprietary Tooling

Key Skills

Software

Internal/Proprietary Tooling

Top Subject Matter

Frontier AI Models
Computer-use Evaluation
Web Search Tasks

Top Data Types

TextText
ImageImage

Top Task Types

RLHF
Prompt Response Writing SFT
Evaluation Rating
Function Calling
Fine Tuning

Freelancer Overview

I work as an LLM Trainer/Delivery Data Analyst, contributing data that AI labs use to train their frontier models. My work spans several key areas of model improvement. In RLHF for computer use, I evaluate how frontier AI models execute tasks based on our prompts, rating their performance against lab-provided rubrics. I've done similar evaluation work for web search flows, assessing model-generated search sequences against defined quality criteria. A significant part of my role involves Supervised Fine-Tuning (SFT), where I craft multi-turn conversations between a simulated human and an AI. This includes writing the model's internal reasoning, executing tool calls, and determining follow-up actions based on tool outputs. I've applied this across both tool-calling SFT and agentic SFT reasoning projects, where I construct end-to-end conversations that demonstrate how an AI should select tools, reason through each step, and respond to user prompts. I've also contributed to GUI annotation projects using PyAutoGUI on virtual machines, where I design tasks and record complete interaction workflows. These recordings, which capture every action along with its reasoning and intent, are used to train frontier models on computer-use capabilities.

IntermediateTamilEnglish

Labeling Experience

Data Labeler – GUI Workflow Annotation (LLM Training)

Text
I designed and recorded detailed GUI interaction workflows for training AI models on computer-use tasks. Leveraging tools like PyAutoGUI in virtual environments, I annotated each user action with corresponding intent and reasoning. These annotated workflows supported model learning for stepwise navigation and command execution. • Created end-to-end GUI recordings with action labeling • Documented intent behind each click and keyboard action • Used PyAutoGUI and virtual machines to simulate user behavior • Delivered structured data for training agentic computer-use models

I designed and recorded detailed GUI interaction workflows for training AI models on computer-use tasks. Leveraging tools like PyAutoGUI in virtual environments, I annotated each user action with corresponding intent and reasoning. These annotated workflows supported model learning for stepwise navigation and command execution. • Created end-to-end GUI recordings with action labeling • Documented intent behind each click and keyboard action • Used PyAutoGUI and virtual machines to simulate user behavior • Delivered structured data for training agentic computer-use models

2024 - Present

Web Search Sequence Evaluator / AI Model Rater

Text
I conducted assessments of model-generated web search sequences and outputs for quality assurance. Using detailed rubrics, I rated accuracy, relevance, and step-wise reasoning of AI-driven search tasks. My evaluation work directly informed model retraining cycles. • Scored AI web search responses for correctness • Followed standard criteria to ensure objective evaluation • Identified patterns in model errors and inconsistencies • Provided structured feedback to enhance AI model search quality

I conducted assessments of model-generated web search sequences and outputs for quality assurance. Using detailed rubrics, I rated accuracy, relevance, and step-wise reasoning of AI-driven search tasks. My evaluation work directly informed model retraining cycles. • Scored AI web search responses for correctness • Followed standard criteria to ensure objective evaluation • Identified patterns in model errors and inconsistencies • Provided structured feedback to enhance AI model search quality

2024 - Present

SFT Dialog Writer / LLM Data Annotator

TextPrompt Response Writing SFT
I developed multi-turn conversations and reasoning chains for supervised fine-tuning of language models. This included crafting tool-use scenarios, prompt responses, and follow-up actions to improve model agentic capabilities. My work helped shape model behavior through high-quality simulated dialogues. • Authored SFT conversations with step-by-step tool selection and logic • Demonstrated prompt engineering and action logic in annotated dialogues • Focused on realistic human-computer interaction cases • Supported agentic training for computer-use language models

I developed multi-turn conversations and reasoning chains for supervised fine-tuning of language models. This included crafting tool-use scenarios, prompt responses, and follow-up actions to improve model agentic capabilities. My work helped shape model behavior through high-quality simulated dialogues. • Authored SFT conversations with step-by-step tool selection and logic • Demonstrated prompt engineering and action logic in annotated dialogues • Focused on realistic human-computer interaction cases • Supported agentic training for computer-use language models

2024 - Present

LLM Trainer / Delivery Data Analyst (Data Labeling & RLHF Evaluation)

TextRLHF
I contributed to the training of frontier AI models by evaluating their responses on computer-use tasks against specific rubrics. My role involved both hands-on evaluation work and team supervision within RLHF workflows. I ensured annotation quality and consistency through regular review and feedback. • Evaluated AI model outputs during RLHF workflows for computer-use cases • Scored model-generated web search sequences and actions against lab rubrics • Conducted quality reviews for conversation and reasoning chain annotations • Designed feedback processes to maintain annotation accuracy and standards

I contributed to the training of frontier AI models by evaluating their responses on computer-use tasks against specific rubrics. My role involved both hands-on evaluation work and team supervision within RLHF workflows. I ensured annotation quality and consistency through regular review and feedback. • Evaluated AI model outputs during RLHF workflows for computer-use cases • Scored model-generated web search sequences and actions against lab rubrics • Conducted quality reviews for conversation and reasoning chain annotations • Designed feedback processes to maintain annotation accuracy and standards

2024 - Present

Education

K

Karpagam Academy of Higher Education

Bachelor of Engineering, Computer Science and Design

Bachelor of Engineering
2021 - 2025
V

Velammal Bodhi Campus

Higher Secondary Certificate, Science

Higher Secondary Certificate
2019 - 2021

Work History

T

Turing (

Delivery Data Analyst

Location not specified
2025 - Present
T

Turing (

Delivery Data Analyst Intern

Location not specified
2024 - 2025