Andrés Q.

Key Skills

Software

Scale AI

Surge AI

Other

Top Subject Matter

Python programming

code generation

Debugging Domain Expertise

Top Data Types

Text

Image

Document

Top Task Types

RLHF

Evaluation/Rating

Question Answering

Text Generation

Prompt + Response Writing (SFT)

Freelancer Overview

Python Code Generation & Debugging – AI Training Data (Outlier / BlackBeard) RLHF Data Trainer. Brings 6+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include Other. Education includes Bachelor of Science, Universidad Nacional de Colombia (2019) and Certificate, BNC Colombo Americano (2018). AI-training focus includes data types such as Computer Code, Programming, and Text and labeling workflows including RLHF, Evaluation, and Rating.

Labeling Experience

RLHF Response Evaluation – STEM & Mathematics (BlackBeard / Outlier)

OtherTextRLHF

Evaluated and ranked AI-generated responses for a large-scale RLHF training project (BlackBeard) on the Outlier platform, specializing in STEM domains including mathematics (calculus, linear algebra, discrete math), physics, electronics, and engineering. Assessed response pairs across five structured dimensions: Instruction Following, Truthfulness, Verbosity, Prompt Correctness, and Writing Style & Tone. Applied Likert-scale preference rankings with evidence-based written justifications, following strict editorial and naming conventions defined by the project's style guide. Evaluated multimodal tasks combining text and image prompts as part of the project's most recent update. Consistently audited LaTeX syntax accuracy, distinguishing model errors from platform rendering bugs per official guidelines.

2026 - Present

Mathematical LaTeX Correction & Notation Standardization

TextText Generation

Identified and corrected malformed LaTeX expressions, non-standard mathematical notation, and broken markdown structure in AI-generated STEM responses. Applied KaTeX compatibility standards (dollar-sign delimiters, standard command set) to ensure correct rendering. Flagged and removed random tokens, language switches, and formatting artifacts, produced clean, publication-ready rewritten responses suitable for fine-tuning datasets.

2025 - Present

AI Response Quality Evaluation – Comparative Annotation (Outlier / BlackBeard)

OtherTextEvaluation/Rating

Performed side-by-side comparative evaluation of AI model response pairs within the BlackBeard RLHF pipeline. Scored each response independently across five quality dimensions and produced a final Likert preference ranking (7-point scale) with detailed written justification. Maintained strict consistency standards: uniform criteria across both responses, no AI-assisted evaluation, adherence to naming conventions, and avoidance of linter-flagged contradictions in justification language. Tasks spanned diverse domains, including advanced mathematics, physics, engineering, finance, legal, and healthcare.

2025 - Present

Python Code Generation & Debugging – AI Training Data (Outlier / BlackBeard) RLHF Data Trainer

OtherRLHF

As an RLHF data trainer, I created, reviewed, and debugged Python code responses to AI prompts for training large language models on the Outlier Platform. I evaluated AI-generated Python code for correctness, efficiency, readability, and PEP 8 compliance, providing structured, reference-quality feedback. My tasks included generating step-by-step written justifications, translating debugging insight into RLHF feedback, and validating code behavior in sandboxed environments. • Wrote and reviewed Python responses for coding prompts such as data structures and algorithms • Evaluated output using manual and code execution in platforms like Google Colab • Produced clear written justifications for rankings and feedback • Ensured solutions met rigorous coding and editorial standards

2023 - Present

Education

Work History