Looking for a job

Karanja User

@karanjauser1

Data Annotator and AI Trainer at Atlas Capture, producing model-ready image, video, audio, and text annotations with over 96% accuracy.

Kenya

Message

At Atlas Capture, I annotate image, video, audio, and text datasets, applying ontology guidelines to produce consistent, model-ready data. I also review peer annotations and correct errors, with accuracy rates above 96%.

At Outlier AI, I evaluate AI-generated Kiswahili text, write and paraphrase training examples, and transcribe audio for speech recognition. I provide feedback on model failures to guide dataset improvements and fine-tuning.

My work with Data Annotation Tech and Sigma AI included NLP and conversational data, Kiswahili language contributions, and dataset QA. I bring a linguist’s precision to annotation across text, image, audio, and video.

Experience

Work history, roles, and key accomplishments

Current

Data Annotator & AI Training Specialist

Current

Atlas Capture

Jan 2023 - Present (3 years 9 months)

Annotated diverse datasets including images, video frames, audio clips, and text documents. Performed bounding box drawing, polygon segmentation, semantic labelling, keypoint detection, and entity tagging, while maintaining annotation accuracy above 96%.

Bounding Boxes Polygon Segmentation Semantic Annotation Keypoint Detection Entity Tagging Quality Assurance Ontology Adherence Data Privacy

Current

Kiswahili AI Annotator & Transcriptionist

Current

Outlier AI

Jan 2022 - Present (4 years 9 months)

Evaluated and rated AI-generated Kiswahili text outputs for fluency, grammatical accuracy, cultural relevance, and factual correctness. Wrote and paraphrased Kiswahili prompts and transcribed audio recordings for speech recognition model training.

Kiswahili AI Evaluation Transcription LLM NLP Quality Assurance

Data Annotation Technician

Data Annotation Tech

Jan 2022 - Dec 2023 (1 year 11 months)

Completed text annotation tasks including sentiment labelling, intent classification, named entity recognition, and coreference resolution for NLP model training. Reviewed and corrected AI-generated outputs for factual accuracy and safety compliance.

Sentiment Labeling Intent Classification Named Entity Recognition Coreference Resolution NLP AI Output Evaluation Quality Assurance

Kiswahili Language Contributor

Sigma AI

Jan 2021 - Dec 2022 (1 year 11 months)

Produced original Kiswahili sentence pairs, translations, and conversational data for NLP research pipelines. Conducted QA reviews on peer-contributed datasets and performed audio-to-text transcription.

Kiswahili Translation Transcription NLP Quality Assurance