Michael B.

Key Skills

Software

Don't disclose

Label Studio

Top Subject Matter

Computational linguistics and annotation for NLP datasets

Multilingual counterspeech dataset construction for NLP

Top Data Types

Text

Audio

Top Task Types

Data Collection

Transcription

Classification

Computer Programming/Coding

Freelancer Overview

Computational Linguistics Research Assistant, Dr. Liu’s Computational Linguistics Lab (Enterprise labeling server setup . Brings 7+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Human Signal and Don't disclose. Education includes Bachelor of Science, University of Florida (2026). AI-training focus includes data types such as Computer Code, Programming, and Text and labeling workflows including Evaluation, Rating, and Data Collection.

Labeling Experience

Computational Linguistics Research Assistant, Dr. Liu’s Computational Linguistics Lab (Enterprise labeling server setup and annotation interface)

Built and supported enterprise-grade labeling server workflows via a Human Signal partnership to enable a team of annotators. Developed and maintained an interactive JavaScript annotation interface with a Cloudflare-based backend to streamline per-item labeling operations. Automated supporting pipelines using Python scripts and maintained a SQL database to sustain labeling system uptime. • Secured labeling infrastructure for annotators • Reduced per-item labeling time by 40% via UI/backend improvements • Maintained 98% operational uptime through automation and database maintenance • Supported labeling program performance and reliability

2025 - 2026

Co-authored PANDA dataset work: Paired Anti-hate Narratives Dataset from Asia (dataset generation and curation)

Don't discloseTextData Collection

Led creation of a high-quality Mandarin Chinese hate speech–counterspeech corpus using a data flywheel that combined simulated annealing, LLM-based generation, and human curation. Coordinated a team to design the pipeline for dataset generation and iterative improvement. Analyzed dataset performance and biases with attention to LLM-as-a-Judge evaluation discrepancies. • Curated and generated training/evaluation text pairs for hate speech counterspeech • Designed an end-to-end data flywheel pipeline for dataset quality • Performed bias and performance analysis of generated dataset • Investigated limitations in automated judging metrics

2025 - 2025

Education

Work History