Livingstone Mumelo

Key Skills

Software

Google Cloud Vertex AI

Toloka

Other

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Document

Text

Top Label Types

Text Generation

Text Summarization

Data Collection

Classification

Bounding Box

Polygon

Point Key Point

Geocoding

Relationship

Tracking

Freelancer Overview

I am an AI and data specialist with over 7 years of experience designing, managing, and delivering data-driven solutions in humanitarian, health, and development sectors. My background includes hands-on work in data labeling, annotation, and preparation of high-quality training datasets for AI/ML projects, with a strong focus on ensuring data integrity and alignment with organizational goals. I have developed and maintained data pipelines, ETL processes, and data models using Python, R, SQL, and tools like Power BI, Tableau, and Looker, supporting both structured and unstructured data in domains such as medical research, food security, and agronomy. My experience spans building automated data validation systems, leading annotation teams, and translating complex data into actionable insights for stakeholders. I am adept at bridging technical and non-technical teams, facilitating data governance, and ensuring best practices in data quality, privacy, and documentation throughout the AI training data lifecycle.

Labeling Experience

Geospatial Labeling: Gauge Points + Catchment Polygons for Hydrology Modeling

OtherGeospatial Tiled ImageryBounding BoxPolygon

Prepared geospatial labels required for hydrology pipelines by mapping physical gauges to correct locations and generating/validating catchment polygons. Cleaned latitude/longitude inputs, harmonized CRS, checked topology (self-intersections, gaps), and tagged each geometry with standardized identifiers for downstream modeling. Enforced QA including boundary checks, geometry validity checks, and consistency between gauge naming conventions and spatial features.

Prepared geospatial labels required for hydrology pipelines by mapping physical gauges to correct locations and generating/validating catchment polygons. Cleaned latitude/longitude inputs, harmonized CRS, checked topology (self-intersections, gaps), and tagged each geometry with standardized identifiers for downstream modeling. Enforced QA including boundary checks, geometry validity checks, and consistency between gauge naming conventions and spatial features.

2023

Perception & Security Survey Coding for Peacebuilding

Google Cloud Vertex AITextText GenerationText Summarization

Annotated and coded large-scale perception and KII survey data to support peacebuilding and inclusive development programming. Created a labeling scheme for themes such as trust in institutions, security perceptions, service access, social cohesion, conflict drivers, and accountability. Standardized open-ended responses into structured categories, applied consistent codebooks, validated inter-coder consistency during quality checks, and enforced QA rules (missingness checks, range checks, skip logic validation, duplicate detection). Produced clean, analysis-ready datasets and indicator-ready outputs for reporting.

Annotated and coded large-scale perception and KII survey data to support peacebuilding and inclusive development programming. Created a labeling scheme for themes such as trust in institutions, security perceptions, service access, social cohesion, conflict drivers, and accountability. Standardized open-ended responses into structured categories, applied consistent codebooks, validated inter-coder consistency during quality checks, and enforced QA rules (missingness checks, range checks, skip logic validation, duplicate detection). Produced clean, analysis-ready datasets and indicator-ready outputs for reporting.

2023 - 2025

Dataset Sensitivity & PII Tagging for Data Sharing Framework

TolokaDocumentClassification

Labeled datasets and metadata to enable safe data sharing across partners, aligning to data protection requirements. Tagged fields and datasets by sensitivity level (public/internal/restricted), identified PII and quasi-identifiers, and applied rules for minimization, aggregation, anonymization, and access control. Built a practical “label-first” approach to governance so stakeholders could share responsibly while keeping operational value.

Labeled datasets and metadata to enable safe data sharing across partners, aligning to data protection requirements. Tagged fields and datasets by sensitivity level (public/internal/restricted), identified PII and quasi-identifiers, and applied rules for minimization, aggregation, anonymization, and access control. Built a practical “label-first” approach to governance so stakeholders could share responsibly while keeping operational value.

2020 - 2025

Operational Traceability Labeling for Lot Tracking & QA Parameters (Food/Dairy Process)

LabelboxTextRelationshipClassification

Designed and labeled operational datasets for traceability across production stages (intake → pooling → separation → remix → work order). Created consistent labeling for lot codes, transformations, product families, and QA parameters by sub-stage. Implemented validation rules to prevent duplicates and ensure referential integrity across linked records. Built “operator-friendly” labeling logic so front-end users could reliably select the correct lots available for the day while maintaining full lineage.

Designed and labeled operational datasets for traceability across production stages (intake → pooling → separation → remix → work order). Created consistent labeling for lot codes, transformations, product families, and QA parameters by sub-stage. Implemented validation rules to prevent duplicates and ensure referential integrity across linked records. Built “operator-friendly” labeling logic so front-end users could reliably select the correct lots available for the day while maintaining full lineage.

2020 - 2024

Education

Work History