Filipe Oliveira

Key Skills

Software

Appen

Clickworker

Data Annotation Tech

Mindrift

OneForma

Remotasks

Scale AI

Toloka

Telus

Internal/Proprietary Tooling

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Document

Text

Top Task Types

Classification

Computer Programming Coding

Prompt Response Writing SFT

RLHF

Translation Localization

Freelancer Overview

Motivated Computer Science and Engineering student at Instituto Superior Técnico with hands-on experience in AI model training, Human-in-the-Loop (HITL) evaluation, and large-scale model benchmarking. As an AI Model Trainer, I evaluated over 6,500 AI conducted interviews to detect fraud patterns and improve model reliability, contributing structured feedback to enhance detection performance and robustness. Benchmarked multiple state-of-the-art language models by analyzing Python and Java prompts and responses, assessing truthfulness, logical reasoning, instruction adherence, algorithmic complexity, and writing quality. Additionally, I designed adversarial programming and mathematics prompts to stress-test advanced AI systems, exposing reasoning limitations and challenging current SOTA capabilities. Recognized for strong analytical thinking, adaptability, and attention to detail, I perform effectively in high-volume, deadline-driven environments while maintaining rigorous evaluation standards. Passionate about AI innovation and continuous learning, I combine technical depth with structured reasoning to contribute meaningfully to advanced AI systems and research-driven projects.

Labeling Experience

AI Model Trainer - Fraud Detection on AI Interviews

LabelboxVideoClassificationRLHF

Evaluated 6,500+ AI-conducted interviews to detect fraudulent behavior, improving model fraud detection robustness by systematically identifying failure patterns and edge cases through a Human-in-the-Loop (HITL) framework. Increased model reliability and detection accuracy by performing structured fraud analysis, documenting behavioral indicators, and providing high-quality feedback loops to refine classification outputs.

Evaluated 6,500+ AI-conducted interviews to detect fraudulent behavior, improving model fraud detection robustness by systematically identifying failure patterns and edge cases through a Human-in-the-Loop (HITL) framework. Increased model reliability and detection accuracy by performing structured fraud analysis, documenting behavioral indicators, and providing high-quality feedback loops to refine classification outputs.

2025

AI Model Trainer - Code Validation & Generation

Internal Proprietary ToolingComputer Code ProgrammingClassificationRLHF

Participated in AI training tasks focused on validating and evaluating code snippets generated by LLMs in multiple programming languages, including Python, Java, C++, C#, Go, Swift, and Ruby. Responsibilities included checking for syntax correctness, logical accuracy, best practices, and compliance with task requirements. Also involved in prompt engineering to improve code generation quality. Worked on high volume annotation projects ensuring consistency and accuracy for advanced AI coding assistants.

Participated in AI training tasks focused on validating and evaluating code snippets generated by LLMs in multiple programming languages, including Python, Java, C++, C#, Go, Swift, and Ruby. Responsibilities included checking for syntax correctness, logical accuracy, best practices, and compliance with task requirements. Also involved in prompt engineering to improve code generation quality. Worked on high volume annotation projects ensuring consistency and accuracy for advanced AI coding assistants.

2025

AI Model Trainer - Text Evaluation & Prompt Optimization

Internal Proprietary ToolingTextText GenerationRLHF

Worked on large-scale AI training projects focused on improving LLM performance for Portuguese and English text. Tasks included evaluating AI-generated responses for accuracy, clarity, and relevance, performing fact-checking and intent verification, and optimizing prompts for better model output. Ensured adherence to strict quality guidelines and maintained high accuracy scores across thousands of annotations. The project contributed to refining model behavior for conversational AI systems used by major tech companies.

Worked on large-scale AI training projects focused on improving LLM performance for Portuguese and English text. Tasks included evaluating AI-generated responses for accuracy, clarity, and relevance, performing fact-checking and intent verification, and optimizing prompts for better model output. Ensured adherence to strict quality guidelines and maintained high accuracy scores across thousands of annotations. The project contributed to refining model behavior for conversational AI systems used by major tech companies.

2025

Education

Work History