Kix T.

Key Skills

Software

Don't disclose

Top Subject Matter

Large Language Model (LLM) output evaluation and alignment for technical domains

AI red teaming

security auditing

Top Data Types

Text

Image

Document

Top Task Types

Red Teaming

Question Answering

Text Generation

Text Summarization

Data Collection

Freelancer Overview

AI Output Evaluation & RLHF Feedback Specialist. Core strengths include Don't disclose, Kali Linux, and Internal. AI-training focus includes data types such as Text, Computer Code, and Programming and labeling workflows including Evaluation, Rating, and Red Teaming.

Labeling Experience

Data Preparation via LLM-Assisted Automation & Regex Cleaning

Generated and debugged shell and automation scripts to clean, restructure, and transform raw technical text into usable datasets for AI training and analysis. Built robust regular-expression structures and scripts to sanitize and convert unstructured inputs into structured formats. Collaborated with LLM-assisted workflow automation to support repeatable data preparation steps. • Authored CLI automation scripts with LLM assistance • Designed regex-based cleaning and transformation pipelines • Troubleshot and optimized script logic • Prepared structured datasets from raw technical text

2023 - Present

AI Red Teaming & Safety Auditing Practitioner

TextRed Teaming

Engineered adversarial prompt sequences to test the safety thresholds and vulnerability limits of emerging AI models. Leveraged Kali Linux knowledge to design jailbreak-style scenarios and evaluate model defenses. Documented and refined red-team prompt strategies to stress-test compliance and robustness. • Created adversarial/jailbreak prompt cases • Tested model safety guardrails and failure modes • Iterated prompts based on observed model behavior • Conducted security-focused evaluation of LLM responses

2023 - Present

AI Output Evaluation & RLHF Feedback Specialist

Don't discloseText

Evaluated LLM outputs by analyzing and rating responses for accuracy, logic, and factual consistency to support alignment and behavioral guardrails. Produced precise human feedback for fine-tuning and reinforcement learning workflows. Performed systematic review of model responses to ensure technical correctness for engineering-oriented tasks. • Rated response quality across multiple criteria • Wrote human feedback for model behavior adjustment • Validated factual and logical consistency in outputs • Supported RLHF-style evaluation cycles

2023 - Present

Education

Work History