Kimani

I am a highly disciplined and detail-oriented AI Evaluation and Machine Learning Specialist with over six years of experience optimizing deep learning models and multimodal computer vision systems through rigorous data annotation. Expert in precise behavioral segmentation, visual-to-text grounding, and high-level captioning under strict objective guidelines. Proven track record of maintaining an 85–90%+ accuracy benchmark in high-velocity environments, utilizing advanced keyboard shortcut workflows and custom macros to meet demanding operational efficiency targets.

Adept at navigating structured, tracked environments without compromising quality, combining machine learning frameworks with deep technical rigor to ensure 100% data fidelity for model training and verification.

Senior AI Evaluation & Machine Learning Annotator at Outlier / Scale AI (2024-01 – Present)

• Execute high-level video captioning and action labeling for advanced multimodal machine learning models, consistently hitting an internal quality benchmark score of 93% accuracy.
• Segment dense video timelines into precise, action-based discrete windows, ensuring absolute temporal continuity with zero overlapping frames or unassigned gaps.
• Formulate objective, 1–2 sentence summaries of complex visual sequences in plain English, strictly stripping away human projections, body-part references, or cognitive intent assumptions.
• Conduct comprehensive QA reviews on model-generated metadata, editing automated text strings into perfect compliance with strict structural and formatting guidelines.
• Maintain an operational speed of processing 1-minute video pipelines within a strict 4-minute limit, utilizing customized platform macros and advanced system hotkeys to maximize daily throughput.

Machine Learning Data Annotator & 3D Specialist at CloudFactory (2022-11 – 2023-12)

• Performed complex 2D Bounding Box Labeling and 3D Data Annotation on dynamic video datasets for autonomous systems and behavioral machine learning models.
• Reviewed and corrected automatically generated automated captions, identifying pixel-level boundary errors and contextual alignment mismatches.
• Operated continuously within strict, time-tracked, screen-monitored environments, maintaining an average quality rating of 91% while adhering to a strict "no tab switching" focus policy.
• Ran background screen recording software and custom diagnostic tools simultaneously without sacrificing system performance or task velocity.

Machine Learning Data Trainer & Content Annotator at Tech-Synergy (2021-06 – 2022-10)

• Annotated massive text and video datasets for localized semantic variation, focusing on structural consistency and tense standardization (simple present tense enforcement) for NLP and vision models.
• Audited automatically generated machine learning translations and visual descriptions, flagging hallucinated details and correcting contextual alignment errors.
• Collaborated closely with international engineering teams to iterate on machine learning annotation instruction manuals and edge-case reference sheets.

Junior Machine Learning Analyst & Systems Assistant at Samasource (2020-01 – 2021-05)

• Handled large-scale data pre-processing and text-to-image alignment for machine learning classification and computer vision models.
• Leveraged machine learning academic training to build custom hotkey configurations, drastically reducing manual task completion times across the immediate team.
• Audited training datasets for strict compliance with objective client guidelines, systematically filtering out subjective user assumptions and intent projections.

Bachelor of Science in Machine Learning – Kenyatta University (2019)