Dennis Muriuki

Key Skills

Software

Scale AI

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Text

Video

Top Label Types

Action Recognition

Classification

Freelancer Overview

I specialize in high-precision data labeling, annotation, and AI training data quality for complex domains, with hands-on experience advancing to Tier 2 on the Atlas Capture platform through rigorous accuracy assessments. My background combines deep understanding of large language model (LLM) reasoning, prompt engineering, and systematic error identification, particularly in medical, pharmaceutical, and healthcare datasets where domain expertise is critical. I’ve led projects that improved dataset quality, automated support workflows, and directly enhanced AI model reliability by applying clinical-grade verification and structured data delivery. I thrive in remote, asynchronous environments, consistently delivering production-standard results in LLM evaluation, human action annotation, and structured data normalization, and bring a strong track record of translating complex requirements into actionable, high-quality AI training data.

Labeling Experience

LLM Output Evaluation & Healthcare Domain Annotation — Atlas Capture Platform

Scale AIVideoAction Recognition

Worked as a Tier 2 qualified contributor on the Atlas Capture platform, advancing beyond Tier 1 after passing all required accuracy assessments — a competitive upgrade that a significant proportion of contributors do not achieve. Core work involved evaluating large language model outputs for factual accuracy, identifying hallucinations and confident but incorrect reasoning, and flagging errors before they could propagate into training data or downstream AI applications. Applied pharmaceutical and clinical expertise (ACLS, PALS, BLS certified) to healthcare-specific annotation tasks, providing subject matter accuracy that general contributors cannot replicate. This included verifying medical claims against clinical ground truth, assessing drug-related AI outputs for accuracy, and ensuring health-related training data met the precision standards required for medical AI deployment.

Worked as a Tier 2 qualified contributor on the Atlas Capture platform, advancing beyond Tier 1 after passing all required accuracy assessments — a competitive upgrade that a significant proportion of contributors do not achieve. Core work involved evaluating large language model outputs for factual accuracy, identifying hallucinations and confident but incorrect reasoning, and flagging errors before they could propagate into training data or downstream AI applications. Applied pharmaceutical and clinical expertise (ACLS, PALS, BLS certified) to healthcare-specific annotation tasks, providing subject matter accuracy that general contributors cannot replicate. This included verifying medical claims against clinical ground truth, assessing drug-related AI outputs for accuracy, and ensuring health-related training data met the precision standards required for medical AI deployment.

2023

Data Capture, Verification & Structured Data Delivery — KNEC National Contract

Internal Proprietary ToolingTextClassification

Selected by the Kenya National Examinations Council (KNEC) for a national-level data capture contract, operating under strict government-grade institutional accuracy standards. The role required normalising inconsistent, high-volume source records into clean, structured digital outputs delivered to precise specification — with zero tolerance for error given the institutional consequences of inaccurate examination data. Delivered consistent, verified output across a compressed two-week high-volume contract, demonstrating the ability to sustain accuracy under significant time pressure across diverse and varied input types. Every record was cross-checked before submission, applying the same multi-source verification discipline developed through years of pharmaceutical prescription checking and clinical documentation.

Selected by the Kenya National Examinations Council (KNEC) for a national-level data capture contract, operating under strict government-grade institutional accuracy standards. The role required normalising inconsistent, high-volume source records into clean, structured digital outputs delivered to precise specification — with zero tolerance for error given the institutional consequences of inaccurate examination data. Delivered consistent, verified output across a compressed two-week high-volume contract, demonstrating the ability to sustain accuracy under significant time pressure across diverse and varied input types. Every record was cross-checked before submission, applying the same multi-source verification discipline developed through years of pharmaceutical prescription checking and clinical documentation.

2025 - 2025

Education

Work History