Cristian Yepes

Key Skills

Software

CVAT

Data Annotation Tech

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Computer Code Programming

Image

Top Task Types

Audio Recording

Computer Programming Coding

Prompt Response Writing SFT

RLHF

Freelancer Overview

Specialized in RLHF and multi-modal AI training, leveraging a software engineering background (C++, Python) to evaluate complex code generation tasks. My experience at Outlier focuses on rigorous logic reasoning, debugging, and syntax optimization, ensuring LLMs produce executable and efficient code. I treat data labeling as a precision engineering task, prioritizing logical consistency and accuracy. Beyond code, I possess versatile expertise in Audio (Speech Synthesis/ASR) and Computer Vision. I excel in Prompt Engineering and edge-case management, refining instructions to ensure models deliver contextually accurate, safe, and natural responses. My approach combines technical knowledge with linguistic precision to handle diverse data pipelines effectively.

Labeling Experience

Semantic Segmentation for Industrial QA (Quality Assurance)

Scale AIImageAction RecognitionRLHF

High-precision polygon annotation for surface defect detection in manufacturing datasets. Validated edge cases in variable lighting conditions to reduce false positives in automated inspection models

High-precision polygon annotation for surface defect detection in manufacturing datasets. Validated edge cases in variable lighting conditions to reduce false positives in automated inspection models

2025

Machine Learning

ImagePolygon

Advanced Python developer proficient in BeautifulSoup, Selenium, and Playwright for large-scale data extraction. Expert in bypassing JavaScript, AJAX, and infinite scroll roadblocks while managing complex multi-level site architectures. Skilled in building resilient workflows with Apify, OpenRouter, and AI agents to automate and scale scraping tasks. Proven ability to handle batching and parallelization for high-performance data processing and delivery. Specialized in cleaning and normalizing datasets into structured CSV, JSON, and Google Sheets formats. Strong focus on implementing fail-safe strategies like custom selectors, fallbacks, and retry logic. Fluent in English (B2+) with the technical mindset to document edge cases and collaborate in hybrid AI environments.

Advanced Python developer proficient in BeautifulSoup, Selenium, and Playwright for large-scale data extraction. Expert in bypassing JavaScript, AJAX, and infinite scroll roadblocks while managing complex multi-level site architectures. Skilled in building resilient workflows with Apify, OpenRouter, and AI agents to automate and scale scraping tasks. Proven ability to handle batching and parallelization for high-performance data processing and delivery. Specialized in cleaning and normalizing datasets into structured CSV, JSON, and Google Sheets formats. Strong focus on implementing fail-safe strategies like custom selectors, fallbacks, and retry logic. Fluent in English (B2+) with the technical mindset to document edge cases and collaborate in hybrid AI environments.

2024 - Present

Advanced Logic & Chain-of-Thought (CoT) Verification

Scale AITextClassificationText Generation

Analyzed multi-step reasoning traces for LLMs. Identified logical fallacies, edge cases, and hallucinations in business logic scenarios. Annotated strict logical constraints for systematized output generation.

Analyzed multi-step reasoning traces for LLMs. Identified logical fallacies, edge cases, and hallucinations in business logic scenarios. Annotated strict logical constraints for systematized output generation.

2025 - 2025

RLHF for Code Generation (Python)

Scale AIComputer Code ProgrammingComputer Programming Coding

Evaluated and rewrote AI-generated code snippets for efficiency, security, and syntax correctness. Created "Golden Standard" responses for complex algorithmic prompts involving multi-threading and database optimization

Evaluated and rewrote AI-generated code snippets for efficiency, security, and syntax correctness. Created "Golden Standard" responses for complex algorithmic prompts involving multi-threading and database optimization

2024 - 2025

Education

Work History