Santiago Villalobos-Gonzalez

Key Skills

Software

Internal/Proprietary Tooling

Top Subject Matter

AI/ML - LLM Trainer

Data Sciences - Database Maintenance, Organization, and Cataloging

Finance Algorithm Training - Stock Market Training

Top Data Types

Audio

Text

Computer Code Programming

Top Task Types

Classification

Question Answering

Text Summarization

Transcription

Computer Programming/Coding

Data Collection

Freelancer Overview

I spent part of my career at Apple on AI/ML Data Operations, where the job sat exactly between raw interviews and the models that ship in products. I ran structured interviews, then turned those sessions into training-ready assets: segmenting turns, tagging intent and outcome, and fixing edge cases where everyday speech does not match clean textbook examples. A large slice of that work touched Siri-related speech and dialogue data—when audio or transcripts came through my pipeline, I applied consistent labeling rules and annotations so downstream teams could trust what they were training on. That mix stuck with me: you cannot annotate well without knowing where the conversation came from. Conducting the interviews myself meant I caught ambiguity early—accent drift, overlapping speakers, soft confirmations—and documented it in the labels instead of letting it leak into the model as noise. I am comfortable with guidelines that change as the product shifts, with tooling that is half spreadsheet and half script, and with arguing for quality when speed is the default setting.

Labeling Experience

Education

Work History