Alejandro Lavandeira

Key Skills

Software

Clickworker

CVAT

Google Cloud Vertex AI

Labelbox

LabelImg

Label Studio

OpenCV AI Kit (OAK)

Remotasks

Roboflow

Toloka

VoTT

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Image

Text

Top Task Types

Bounding Box

Classification

Computer Programming Coding

Data Collection

Object Detection

Freelancer Overview

As a recent graduate in Automation and Industrial Robotics, with ongoing studies in Robotic Engineering, I have hands-on experience in the full lifecycle of AI training data, particularly in computer vision and machine learning projects. I have independently labeled and annotated hundreds of images and videos for object detection tasks, using tools like LabelImg and CVAT to create high-quality datasets. This directly fed into training YOLO models from scratch, where I managed the entire process—from data collection and annotation to model fine-tuning and evaluation—achieving improved accuracy in real-world industrial applications such as robotic assembly line monitoring.What sets me apart is my end-to-end expertise across ML paradigms, including reinforcement learning (RL) pipelines built from zero using frameworks like Stable Baselines, and A/B testing for large language models like Gemini to refine response quality. I excel at validating model outputs for factual accuracy, identifying biases, and ensuring dataset diversity, which has honed my attention to detail and critical thinking. This blend of technical proficiency in vision tasks, ethical data handling, and practical project execution makes me a versatile contributor to AI training initiatives.

Labeling Experience

Games data annotation and train yolo models

CVATImageBounding BoxObject Detection

I have manually annotated thousands of images, either independently or with assistance from models I trained using my own annotated datasets—all under my direct supervision. These annotations supported multiple video game projects, where the goal was to detect all in-game objects to enable automated gameplay and winning strategies. This involved image annotation, object detection techniques, and color masking to generate high-quality labeled datasets, which I then used to train several YOLOv11 models tailored to each video game. For each model, I handled the full deployment process, ensuring seamless and accurate functionality in real-time applications.

I have manually annotated thousands of images, either independently or with assistance from models I trained using my own annotated datasets—all under my direct supervision. These annotations supported multiple video game projects, where the goal was to detect all in-game objects to enable automated gameplay and winning strategies. This involved image annotation, object detection techniques, and color masking to generate high-quality labeled datasets, which I then used to train several YOLOv11 models tailored to each video game. For each model, I handled the full deployment process, ensuring seamless and accurate functionality in real-time applications.

2025 - 2025

Education

Work History