Scott Nelson

Meta, Remote — Linguistic Data Analyst

AUGUST 2023 - AUGUST 2024

• Collaborated with cross-functional teams (Engineering, Subject Experts, Product, Policy, and Research) to fine‑tune LLM training datasets, ensuring safety and bias alignment in Meta AI’s Llama 2 & Llama 3 projects before deployment.
• Conducted thorough research into state of the art techniques and methods for LLM development, red-teaming, machine learning optimization of prompting, and natural language processing metrics.
• Participated in daily, collaborative meetings for brainstorming experimentation, communicating insights, and troubleshooting.
• Stress test LLM to determine areas of weakness in model safety and analyze language generation. Achieving a 65% reduction in violations—with up to 100% reduction in critical hazard categories—ensuring rigorous safety and bias alignment pre—deployment.
• Compile training datasets tailored to fine tuning needs
• Develop across the board metrics and standards for model training.
• Communicate linguistic insights accessibly and succinctly, while considering technical and computational constraints.

Loyola Marymount University, Los Angeles — Research Assistant

NOVEMBER 2021 - SEPTEMBER 2023

• Conducted advanced statistical analyses from electroencephalogram (EEG) data collection and to support neuroscience research, transforming raw time series data into meaningful comparisons with abnormal neuropsychological research.
• Collaborated with interdisciplinary teams to review literature and implement novel testing designs that bridged cognitive science with emerging AI theory. Conducted critical reviews of research methods for cognitive science.
• Assisted in the development and validation of ensemble transformer architecture. Combining decision-tree and long context transformer architectures creating an ensemble pipeline to predict readers’ subjective perceptions of bias.

Tenant’s Law Firm, Los Angeles — Internship

MAY 2022 - DECEMBER 2022

• Automated data‐intelligence reporting by integrating APIs and JSON data pipelines with the Customer Relationship Management (CRM) system, using a scheduled workflow automation tool (n8n) to prioritize workload for attorneys
• Integrated a document drafting software to cut routine legal drafting time by 80% via digitization and automation.
• Ordered public records requests from the Los Angeles Metro area’s multiple municipal bodies; client/customer service.

Loyola Marymount University, Los Angeles, CA — Bachelor of Arts in Psychology

AUGUST 2019 - MAY 2023

Minors: Computer Science; Statistics & Data Science

Undergrad Coursework: Cognitive Neuroscience • Behavioral Neuroscience • Artificial Intelligence • Programming Abstractions • Data Structures • Research Methods • Multivariate Statistics • Applied Linear Algebra • Probability & Statistics

Grad-level Coursework: Data Science for Machine Learning • Cognitive Systems Design • Natural Language Processing • Agile Project Management