Key Skills

Software

AWS SageMaker

Anno-Mage

CVAT

Doccano

Labelbox

Label Studio

LightTag

SuperAnnotate

Top Subject Matter

Object Detection and Classification for Autonomous Vehicles

NLP Model Training for Customer Support Chatbots

Data Labeling for AI in Healthcare (e.g., combining text and image data)

Top Data Types

Audio

Image

Text

Top Task Types

Classification

Computer Programming Coding

Fine Tuning

Object Detection

Text Summarization

Freelancer Overview

I’m an experienced AI Engineer with over 8 years of hands-on experience specializing in object detection, audio analysis, and deepfake technology. My work focuses on developing, training, and deploying AI models using frameworks like TensorFlow and PyTorch to solve complex problems and deliver real-world solutions. In object detection, I’ve worked on projects such as creating computer vision systems for autonomous vehicles and real-time surveillance. I’ve fine-tuned models like YOLO and Faster R-CNN for tasks such as detecting pedestrians, recognizing traffic signs, and tracking anomalies. I’ve also managed the entire pipeline, from designing annotation workflows to processing large datasets, ensuring high accuracy and efficiency in detection models. For audio analysis, I’ve built AI systems capable of handling speech recognition, emotion detection, and sound classification. By using tools like Librosa and DeepSpeech, I’ve developed models that extract features from noisy audio and enable applications such as sentiment analysis, transcription, and industrial safety monitoring. I’ve also refined audio datasets by implementing preprocessing techniques like noise reduction, segmentation, and spectrogram generation.

Labeling Experience

Image and behavious detection in construction site

LabelboxVideoObject DetectionAction Recognition

This project involved developing and labeling video data to train AI models for monitoring safety and behavior in construction sites. The goal was to identify unsafe actions, track worker movements, and detect potential hazards such as improper use of equipment or absence of safety gear. Using Labelbox, I annotated large-scale video datasets with bounding boxes, action labels, and tracking information to ensure accurate detection of critical behaviors. The project covered labeling over 500 hours of video footage, ensuring consistency and high-quality annotations. I also fine-tuned object detection models like YOLO and Faster R-CNN for worker and equipment identification, and integrated action recognition frameworks to classify behaviors such as lifting, operating machinery, and potential falls or collisions. To maintain quality, strict QA measures were implemented, including inter-annotator agreement checks and automated validation processes.

This project involved developing and labeling video data to train AI models for monitoring safety and behavior in construction sites. The goal was to identify unsafe actions, track worker movements, and detect potential hazards such as improper use of equipment or absence of safety gear. Using Labelbox, I annotated large-scale video datasets with bounding boxes, action labels, and tracking information to ensure accurate detection of critical behaviors. The project covered labeling over 500 hours of video footage, ensuring consistency and high-quality annotations. I also fine-tuned object detection models like YOLO and Faster R-CNN for worker and equipment identification, and integrated action recognition frameworks to classify behaviors such as lifting, operating machinery, and potential falls or collisions. To maintain quality, strict QA measures were implemented, including inter-annotator agreement checks and automated validation processes.

2022 - 2024

Education

Work History