Dessalegn Dagnaw

Key Skills

Software

Appen

Clickworker

CVAT

Data Annotation Tech

Mindrift

OneForma

Remotasks

Toloka

Telus

Scale AI

Top Subject Matter

Computer Vision (Image & Video Annotation, Object Detection, Segmentation)

Natural Language Processing (Text Annotation, Sentiment Analysis, Named Entity Recognition)

Autonomous Vehicles (Sensor Data Annotation, LiDAR, Radar, and Camera Data)

Top Data Types

Image

Text

Video

Top Task Types

Bounding Box

Data Collection

Polygon

Segmentation

Text Generation

Freelancer Overview

I am a highly experienced AI Data Labeling Specialist with over 8 years of experience in annotating and maintaining large datasets to machine learning applications, especially computer vision to NLP. Some of my major areas of specializations include object detection, image segmentation, sentiment analysis, and named entity recognition using data labeling. Extensive work was done with tools like Labelbox, Scale AI, and CVAT in ensuring the quality of the datasets and that they fall in line with the project objectives. Notable projects include leading data annotation for autonomous vehicle systems, whereby I've annotated thousands of images and videos to increase the accuracy of object detection. I will handle NLP tasks for retail companies in improving their insight into customer engagement by text classification and sentiment analysis. This includes technical expertise in AI/ML frameworks such as TensorFlow and PyTorch, combined with workflow optimization and collaboration across diverse teams, hence making me unique for the efficient and effective delivery of data labeling solutions.

Labeling Experience

Image annotation

Scale AIImageBounding BoxQuestion Answering

At Remotasks, I worked on a large-scale image annotation project focused on object detection and image segmentation for autonomous vehicle systems. Using bounding boxes and polygon segmentation, I accurately labeled objects such as pedestrians, vehicles, traffic signs, and road infrastructure in thousands of images. These labeled datasets were used to train machine learning models, enhancing their ability to recognize and navigate complex driving environments. My attention to detail ensured the quality and consistency of the annotations, which were critical for improving the model's real-time object detection capabilities, contributing to safer and more reliable autonomous driving systems.

At Remotasks, I worked on a large-scale image annotation project focused on object detection and image segmentation for autonomous vehicle systems. Using bounding boxes and polygon segmentation, I accurately labeled objects such as pedestrians, vehicles, traffic signs, and road infrastructure in thousands of images. These labeled datasets were used to train machine learning models, enhancing their ability to recognize and navigate complex driving environments. My attention to detail ensured the quality and consistency of the annotations, which were critical for improving the model's real-time object detection capabilities, contributing to safer and more reliable autonomous driving systems.

2018

Data annotation and labelling

Scale AIVideoBounding BoxPolygon

The project at Scale AI will entail data annotation captured with cameras, LIDAR, and radar sensors mounted on self-driving cars. That annotation, categorically speaking, involves point-to-object labeling in images and video frames, like pedestrians, vehicles, traffic signs, road markings, and other obstacles. It also involves the tasks of 3D bounding box labeling, semantic segmentation, and lane detection, which commonly help most autonomous systems identify real-time objects and understand the environment. The data is curated with extreme care regarding annotation so that the computer vision models can accurately detect objects for the vehicle to make decisions at traffic lights, to avoid obstacles, and to make decisions on complex road situations. Quality assurance, on the other hand, is an important contributor to the project; it involves many steps inclined toward the strict check that the annotations are consistent and as correct for the self-driving system's safety.

The project at Scale AI will entail data annotation captured with cameras, LIDAR, and radar sensors mounted on self-driving cars. That annotation, categorically speaking, involves point-to-object labeling in images and video frames, like pedestrians, vehicles, traffic signs, road markings, and other obstacles. It also involves the tasks of 3D bounding box labeling, semantic segmentation, and lane detection, which commonly help most autonomous systems identify real-time objects and understand the environment. The data is curated with extreme care regarding annotation so that the computer vision models can accurately detect objects for the vehicle to make decisions at traffic lights, to avoid obstacles, and to make decisions on complex road situations. Quality assurance, on the other hand, is an important contributor to the project; it involves many steps inclined toward the strict check that the annotations are consistent and as correct for the self-driving system's safety.

2018

Text generation

MindriftTextText GenerationText Summarization

In this project focused on creating high-quality prompt and response datasets for fine-tuning AI models in conversational AI applications. I crafted detailed prompts reflecting common user queries and generated contextually appropriate responses. These prompts and responses were used to train models for chatbots and virtual assistants, improving their ability to respond accurately and naturally. Additionally, sentiment and intent classification were applied to ensure the AI understood user emotions and context. The project enhanced the model’s performance in customer service and NLP-driven tasks, ensuring more meaningful, relevant interactions.

In this project focused on creating high-quality prompt and response datasets for fine-tuning AI models in conversational AI applications. I crafted detailed prompts reflecting common user queries and generated contextually appropriate responses. These prompts and responses were used to train models for chatbots and virtual assistants, improving their ability to respond accurately and naturally. Additionally, sentiment and intent classification were applied to ensure the AI understood user emotions and context. The project enhanced the model’s performance in customer service and NLP-driven tasks, ensuring more meaningful, relevant interactions.

2024 - 2024

LLM in Machine Learning

AppenTextQuestion AnsweringText Generation

I am working on an Appen data collection project focused on improving AI and machine learning models for NLP and speech recognition systems. This involves collecting text and audio data of diverse nature from different sources in order to create a dataset that covers all the bases during the training of AI models. I engage in data collection, tagging, and organization in various formats: transcription of audio recordings, sentiment tagging of customer feedback, and text classification tasks. My work contributes to improving the quality of NLP models, which will have wider applications in customer service automation and voice-activated technologies. I make sure that the information collected is clean and consistent, according to the quality standards set by the project, which improves the results in real-world applications of AI models.

I am working on an Appen data collection project focused on improving AI and machine learning models for NLP and speech recognition systems. This involves collecting text and audio data of diverse nature from different sources in order to create a dataset that covers all the bases during the training of AI models. I engage in data collection, tagging, and organization in various formats: transcription of audio recordings, sentiment tagging of customer feedback, and text classification tasks. My work contributes to improving the quality of NLP models, which will have wider applications in customer service automation and voice-activated technologies. I make sure that the information collected is clean and consistent, according to the quality standards set by the project, which improves the results in real-world applications of AI models.

2021 - 2024

LLM Machine learning

Scale AITextEntity Ner ClassificationQuestion Answering

In my work with the Prompt + Response Writing project at Remotasks, I focused on crafting prompts and refining AI-generated responses for conversational models. I created clear and relevant prompts to help the AI produce accurate and contextually appropriate replies. Additionally, I reviewed and edited the model's responses to ensure they were coherent and aligned with the intended tone. A key part of my role was identifying areas where the AI could improve and providing feedback to enhance its understanding and performance. I also tested the system’s ability to handle different types of queries, ensuring the generated responses were engaging and accurate. My contributions played a key role in developing smarter AI systems capable of creating more natural, human-like conversations across various applications, improving user interactions.

In my work with the Prompt + Response Writing project at Remotasks, I focused on crafting prompts and refining AI-generated responses for conversational models. I created clear and relevant prompts to help the AI produce accurate and contextually appropriate replies. Additionally, I reviewed and edited the model's responses to ensure they were coherent and aligned with the intended tone. A key part of my role was identifying areas where the AI could improve and providing feedback to enhance its understanding and performance. I also tested the system’s ability to handle different types of queries, ensuring the generated responses were engaging and accurate. My contributions played a key role in developing smarter AI systems capable of creating more natural, human-like conversations across various applications, improving user interactions.

2019 - 2024

Show more

Education

Work History