Helen Satchwell

Strong information technology professional and experienced teacher with a master’s degree focused in Hispanic and Latin American Languages, Literature, and Linguistics from the University of Colorado Boulder. Skilled in Academic Writing, English, Spanish, Content Creation. Unique background as an educator, writer, and linguistics student who has worked as with large language model improvement. I currently work as a machine learning data linguist for a AWS. During my three years with The company, I have gained extensive experience in collecting, analyzing, and visualizing data. I also have intermediate skills in Python, SQL, data modeling, and BI tools. Finally, I have a strong understanding of data quality and accuracy standards.

As a Machine Learning Data Linguist III at Amazon Web Services - Bedrock (Dec 2024 - Current), I benchmark and generate training data across policies such as harmful content filtering, denied topic filtering, and personally identifiable information (PII) filtering, ensuring compliance with data quality standards. I design, categorize, and evaluate prompts to enhance model performance, creating diverse and natural responses aligned with user intent. I conduct adversarial testing by crafting prompts to identify model vulnerabilities, collaborating with engineers to implement system improvements. I collect retrieval data from prompts, analyze factual responses, and determine optimal references for improving model accuracy. Furthermore, I generate training datasets with examples for completing complex user prompts that require interaction with external applications, enhancing real-world usability and train vendors. I also evaluate data collection and annotation processes, providing actionable feedback to improve workflows and tools, and propose tooling enhancements and refine system processes to optimize data generation pipelines.

During my time at Amazon Web Services - Transcribe (May 2022 – Dec 2024), I delivered high-quality training data to improve machine learning system performance, driving data-driven decision-making. I managed operations for a 33-member team, including capacity planning, performance tracking, and standardizing onboarding processes. I developed tools, protocols, and annotation guidelines to streamline data processing and enhance quality. I conducted error trend analyses, created corrective action plans, and collaborated with engineers to refine annotation tools and workflows. I defined and audited data quality metrics to ensure compliance with accuracy standards. I collaborated with team members and stakeholders to improve dashboards to track team performance and quality metrics, improving visibility and decision-making across teams. I also wrote task lead documentation, trained team members, and mentored task leads, fostering a collaborative and knowledge-driven environment.

As an Independent Contractor (2019 - 2024), I performed QA for Spanish-translated software and web applications, ensuring linguistic accuracy and usability. I edited and subtitled videos using Adobe Premiere for focus group studies and researched and authored educational content for homeschool students and teachers on the Study.com platform.

As a Spanish Lecturer at the University of Colorado Denver (2017 - 2018), I taught Beginning Spanish II to undergraduate students, designing curriculum and integrating digital resources to enhance learning. I developed activities, assignments, and assessments promoting real-world application of language skills. 
 

During my time as a Spanish Teaching Assistant at the University of Colorado Boulder (2015 - 2017), I delivered Spanish I & II courses, creating engaging lesson plans and alternate assignments to cater to diverse learning styles and collaborated with fellow teaching assistants to develop innovative teaching strategies and materials.

My education includes an M.A. in Spanish Linguistics and a B.A. in Linguistics from the University of Colorado Boulder, as well as a B.A. in English Pedagogy from Universidad Nacional de Andrés Bello.