Maria Pavlenko

Key Skills

Software

Appen

Top Subject Matter

No subject matter listed

Top Data Types

Document

Text

Top Task Types

Translation Localization

Freelancer Overview

I am a multilingual legal translator and interpreter with over 10 years of experience handling complex, sensitive documents across English, Thai, Spanish, Ukrainian, and Russian. My work requires an exceptional attention to detail, accuracy, and cultural nuance—skills directly applicable to high-quality data labeling and annotation for AI and NLP projects. I am highly proficient with CAT tools such as MemoQ, OmegaT, CafeTran, and SDL Trados, and have experience with localization platforms like Crowdin and Smartcat, as well as AI tools like ChatGPT. My background includes editing, proofreading, and QA, ensuring data consistency and reliability in multilingual environments. I am adept at managing high-volume workflows, collaborating with diverse teams, and delivering precise, well-structured data for training and improving AI systems.

Labeling Experience

AI Language Evaluator | ASR Quality Analyst

AppenAudioEvaluation Rating

As an AI Language Evaluator and ASR Quality Analyst, I conducted extensive evaluation of automatic speech recognition models across diverse and code-switched audio datasets. My work involved reviewing over 1,800 long-form audio segments under strict QA production guidelines, focusing on semantic accuracy and error taxonomy. I provided structured, detailed feedback on model performance and risks, ensuring adherence to annotation guidelines and production standards. • Executed large-scale comparative analysis of ASR outputs for Ukrainian-Russian code-switched content. • Applied WER-informed error taxonomy, assessing for omission, hallucination, and normalization errors. • Identified and documented LLM-induced hallucinations and spurious continuations in transcriptions. • Delivered independent, high-throughput QA with minimal supervision.

As an AI Language Evaluator and ASR Quality Analyst, I conducted extensive evaluation of automatic speech recognition models across diverse and code-switched audio datasets. My work involved reviewing over 1,800 long-form audio segments under strict QA production guidelines, focusing on semantic accuracy and error taxonomy. I provided structured, detailed feedback on model performance and risks, ensuring adherence to annotation guidelines and production standards. • Executed large-scale comparative analysis of ASR outputs for Ukrainian-Russian code-switched content. • Applied WER-informed error taxonomy, assessing for omission, hallucination, and normalization errors. • Identified and documented LLM-induced hallucinations and spurious continuations in transcriptions. • Delivered independent, high-throughput QA with minimal supervision.

2025

Education

Work History