Allan Wachira

Key Skills

Software

CVAT

Data Annotation Tech

Hasty

V7 Labs

Top Subject Matter

"self Driving Car Imagery"

"Road Classification Imagery"

"Satellite image Classification"

Top Data Types

Audio

Image

Text

Top Task Types

Bounding Box

Classification

Cuboid

Polygon

Segmentation

Freelancer Overview

I have extensive experience in data labeling and AI training data, focusing on creating high-quality, annotated datasets to train machine learning models effectively. My expertise includes working with various types of data such as text, images, and audio, and utilizing tools and platforms like Labelbox, Amazon SageMaker Ground Truth, and custom annotation tools. Key projects I've worked on involve developing labeled datasets for image classification, natural language processing, and speech recognition tasks, where I ensured data accuracy and consistency through rigorous validation and quality control processes. In addition to hands-on annotation work, I have experience in designing labeling guidelines and training teams of annotators to maintain high standards across large-scale projects. My skills in data management, attention to detail, and familiarity with different annotation techniques, coupled with my ability to adapt to new technologies, set me apart in the field of AI training data.

Labeling Experience

speech transcription

V7 LabsAudioPoint Key PointSegmentation

This project involved the development of a speech transcription system tailored for the entertainment industry, with a primary focus on converting spoken dialogue from movies, TV shows, interviews, and podcasts into accurate written text. The goal was to enhance accessibility, enable efficient content indexing, and support subtitle generation for multimedia platforms. Key components of the project included: Audio Preprocessing: Implemented noise reduction and speaker diarization to improve transcription quality, especially in dynamic entertainment environments with overlapping dialogues and background sounds. Automatic Speech Recognition (ASR): Utilized state-of-the-art ASR models (such as Whisper or wav2vec 2.0) trained on diverse entertainment datasets to handle various accents, slang, and expressive speech common in media content. Post-processing: Developed algorithms for punctuating, formatting, and segmenting the transcriptions into readable scripts or subtitles.

This project involved the development of a speech transcription system tailored for the entertainment industry, with a primary focus on converting spoken dialogue from movies, TV shows, interviews, and podcasts into accurate written text. The goal was to enhance accessibility, enable efficient content indexing, and support subtitle generation for multimedia platforms. Key components of the project included: Audio Preprocessing: Implemented noise reduction and speaker diarization to improve transcription quality, especially in dynamic entertainment environments with overlapping dialogues and background sounds. Automatic Speech Recognition (ASR): Utilized state-of-the-art ASR models (such as Whisper or wav2vec 2.0) trained on diverse entertainment datasets to handle various accents, slang, and expressive speech common in media content. Post-processing: Developed algorithms for punctuating, formatting, and segmenting the transcriptions into readable scripts or subtitles.

2024 - 2024

Education

Work History