Andre Cival

Key Skills

Software

CVAT

Labelbox

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Image

Video

Audio

Text

Top Label Types

Bounding Box

Polygon

Object Detection

Tracking

Action Recognition

Classification

Emotion Recognition

Audio Recording

Transcription

Question Answering

Text Generation

Text Summarization

RLHF

Prompt Response Writing SFT

Freelancer Overview

I am an experienced AI Training Specialist with over three years of hands-on expertise in data annotation and labeling for machine learning and AI model development. My background spans complex image, video, and audio annotation projects, with a strong focus on computer vision tasks like object detection, semantic segmentation, and advanced video tracking, as well as audio labeling for speech recognition and NLP applications. I am skilled at using tools such as Labelbox, CVAT, Supervisely, and VGG Image Annotator, and have maintained high accuracy standards while working with large-scale datasets. My technical foundation includes Python for data preprocessing and a working knowledge of OpenCV, TensorFlow, and PyTorch. I consistently collaborate with ML engineers to refine annotation guidelines, ensure data quality, and support model training workflows, bringing a detail-oriented and quality-driven approach to every project.

Labeling Experience

Autonomous Vehicle Image Object Detection & Segmentation Project

CVATImageBounding BoxPolygon

Annotated and segmented large-scale image datasets for autonomous vehicle perception systems. Performed high-precision bounding box and polygon annotations for vehicles, pedestrians, traffic signs, lane markings, and roadside objects. Created semantic segmentation masks to improve object recognition accuracy in varying lighting and weather conditions. Collaborated with machine learning engineers to refine annotation guidelines and ensure dataset consistency. Conducted quality assurance reviews to maintain over 98% labeling accuracy. Supported YOLO-based object detection model training by preparing structured datasets optimized for real-time detection performance. Ensured compliance with strict labeling standards and contributed to improving model precision and recall metrics across multiple training cycles.

Annotated and segmented large-scale image datasets for autonomous vehicle perception systems. Performed high-precision bounding box and polygon annotations for vehicles, pedestrians, traffic signs, lane markings, and roadside objects. Created semantic segmentation masks to improve object recognition accuracy in varying lighting and weather conditions. Collaborated with machine learning engineers to refine annotation guidelines and ensure dataset consistency. Conducted quality assurance reviews to maintain over 98% labeling accuracy. Supported YOLO-based object detection model training by preparing structured datasets optimized for real-time detection performance. Ensured compliance with strict labeling standards and contributed to improving model precision and recall metrics across multiple training cycles.

2023

LLM Fine-Tuning & RLHF Data Annotation for Large Language Models

Scale AITextQuestion AnsweringText Generation

Contributed to large-scale fine-tuning and reinforcement learning with human feedback (RLHF) projects for advanced language models. Created high-quality prompt-response pairs for supervised fine-tuning (SFT), ensuring clarity, factual accuracy, and structured reasoning. Performed response evaluation and ranking based on coherence, helpfulness, safety, and alignment guidelines. Conducted red teaming exercises to identify edge cases, bias risks, hallucinations, and policy violations. Labeled datasets for text classification, summarization, and question answering tasks across multiple domains including education, healthcare, and general knowledge. Collaborated with AI research and engineering teams to refine evaluation rubrics and improve model performance across reasoning, safety, and instruction-following benchmarks. Maintained high annotation consistency and adhered strictly to detailed project guidelines and quality standards.

Contributed to large-scale fine-tuning and reinforcement learning with human feedback (RLHF) projects for advanced language models. Created high-quality prompt-response pairs for supervised fine-tuning (SFT), ensuring clarity, factual accuracy, and structured reasoning. Performed response evaluation and ranking based on coherence, helpfulness, safety, and alignment guidelines. Conducted red teaming exercises to identify edge cases, bias risks, hallucinations, and policy violations. Labeled datasets for text classification, summarization, and question answering tasks across multiple domains including education, healthcare, and general knowledge. Collaborated with AI research and engineering teams to refine evaluation rubrics and improve model performance across reasoning, safety, and instruction-following benchmarks. Maintained high annotation consistency and adhered strictly to detailed project guidelines and quality standards.

2024 - 2025

Speech Transcription & Emotion Recognition for Conversational AI

LabelboxAudioClassificationEmotion Recognition

Annotated and processed large-scale speech datasets to support training of conversational AI and speech recognition models. Performed high-accuracy verbatim transcription, speaker identification, timestamp segmentation, and background noise classification across diverse audio environments. Labeled emotional tone categories such as neutral, happy, angry, frustrated, and sad to improve emotion recognition capabilities in voice assistant systems. Conducted waveform analysis, audio cleaning, and segmentation using Audacity and Adobe Audition to ensure high-quality training inputs. Maintained strict quality standards with 98%+ transcription accuracy while following detailed annotation guidelines. Collaborated with NLP and machine learning teams to refine labeling taxonomy and enhance model performance in speech-to-text and sentiment detection tasks.

Annotated and processed large-scale speech datasets to support training of conversational AI and speech recognition models. Performed high-accuracy verbatim transcription, speaker identification, timestamp segmentation, and background noise classification across diverse audio environments. Labeled emotional tone categories such as neutral, happy, angry, frustrated, and sad to improve emotion recognition capabilities in voice assistant systems. Conducted waveform analysis, audio cleaning, and segmentation using Audacity and Adobe Audition to ensure high-quality training inputs. Maintained strict quality standards with 98%+ transcription accuracy while following detailed annotation guidelines. Collaborated with NLP and machine learning teams to refine labeling taxonomy and enhance model performance in speech-to-text and sentiment detection tasks.

2023 - 2024

Video Object Tracking & Action Recognition for Smart Surveillance Systems

LabelboxVideoBounding BoxObject Detection

Annotated large-scale surveillance and real-world motion datasets to train computer vision models for object detection and action recognition. Performed frame-by-frame bounding box labeling and multi-object tracking across continuous video sequences, ensuring consistent object IDs throughout motion cycles. Labeled human activities such as walking, running, loitering, and object interactions to support behavior analysis models. Maintained temporal consistency across thousands of frames while handling occlusions, motion blur, and complex scene transitions. Collaborated with ML engineers to refine tracking protocols and improve model performance metrics, including precision, recall, and mAP scores. Conducted quality assurance audits to maintain high annotation accuracy and ensure compliance with project guidelines.

Annotated large-scale surveillance and real-world motion datasets to train computer vision models for object detection and action recognition. Performed frame-by-frame bounding box labeling and multi-object tracking across continuous video sequences, ensuring consistent object IDs throughout motion cycles. Labeled human activities such as walking, running, loitering, and object interactions to support behavior analysis models. Maintained temporal consistency across thousands of frames while handling occlusions, motion blur, and complex scene transitions. Collaborated with ML engineers to refine tracking protocols and improve model performance metrics, including precision, recall, and mAP scores. Conducted quality assurance audits to maintain high annotation accuracy and ensure compliance with project guidelines.

2022 - 2022

Education

Work History