Brian W.

Key Skills

Software

Other

Don't disclose

Labelbox

Top Subject Matter

Swahili and legal/conversational NLP for RLHF training

Machine translation

clinical speech/phonetic transcription

Top Data Types

Text

Audio

Document

Top Task Types

RLHF

Fine-tuning

Entity (NER) Classification

Red Teaming

Classification

Freelancer Overview

Lead AI Data Specialist, Toptal (Client: Kenyan Legal Aid Commission). Brings 6+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include Other, Don't disclose, and Labelbox. Education includes Doctor of Philosophy, University of Nairobi (2026) and Master of Arts, Kenyatta University (2021). AI-training focus includes data types such as Text and Audio and labeling workflows including RLHF, Fine-tuning, and Entity (NER) Classification.

Labeling Experience

Lead AI Data Specialist - Toptal

LabelboxTextRLHFRed Teaming

Lead cross-functional teams to implement large-scale Reinforcement Learning from Human Feedback (RLHF) workflows for core foundational model training. Drive evaluation and structural validation of multi-turn conversational datasets to improve reasoning, logic patterns, and semantic accuracy. Design Swahili morphosyntax rules, style guides, and taxonomies while performing edge-case analysis and adversarial red-teaming to reduce hallucinations, bias, and validation failures. • Manage remote teams of annotators, developers, and bilingual editors for RLHF execution • Lead dataset evaluation and structural validation to raise performance metrics • Build linguistically grounded Swahili NLP specifications and taxonomic frameworks • Partner with ML engineers to optimize training data pipelines and red-team prompts for safety and quality

2024 - Present

Lead AI Data Specialist, Toptal (Client: Kenyan Legal Aid Commission)

OtherTextRLHF

Implemented large-scale reinforcement learning from human feedback (RLHF) workflows using multi-turn conversational datasets for enterprise model training. Conducted evaluation and structural validation to improve reasoning, logic patterning, and semantic accuracy for dialogue-based tasks. Performed edge-case analysis with adversarial prompting to detect and reduce hallucinations, safety breaches, bias, and syntax errors in training examples. • Managed cross-functional teams of remote annotators, developers, and bilingual editors. • Created Swahili morphosyntax rules, style guides, and taxonomies for specialized NLP pipelines. • Optimized data delivery pipelines with minimal runtime failures due to data validation. • Designed red-teaming/adversarial prompting strategies and recorded findings for remediation.

2024 - Present

Senior Computational Linguist and NLP Training Specialist - Upwork

LabelboxAudio

Provide senior computational linguistics and NLP training expertise through enterprise AI projects, curating and preparing linguistic datasets for model use. Build prompt-response relationships for deep reasoning, logic, coding interpretation, and high-quality text generation. Deliver high-fidelity phonetic and orthographic transcriptions with rigorous QA and tune machine translation alignment between English and Swahili. • Curate, clean, and categorize linguistic datasets for machine translation and clinical speech recognition • Create specialized prompting workflows to activate analytical reasoning and structured generation • Produce consistent medical and technical transcriptions with high QA accuracy • Tune and align machine translations and tokenization parameters while managing multi-month delivery milestones

2022 - 2024

Senior Computational Linguist & NLP Training Specialist, Upwork

Don't discloseTextFine-tuning

Curated, cleaned, and categorized linguistic datasets to support enterprise AI model training, including machine translation and clinical speech recognition. Crafted prompt-response relationships to activate model reasoning, mathematical logic, coding interpretation, and complex text generation behaviors. Completed high-fidelity phonetic and orthographic transcriptions and maintained strong QA with cross-verification for transcription quality. • Tuned thousands of machine translation samples for alignment between human intelligence and model outputs. • Set hard alignment parameters and vocabulary/tokenization mapping between English and Swahili. • Delivered multi-month, end-to-end dataset preparation with aggressive timelines. • Achieved high job success scores across contracts.

2022 - 2024

Language Solutions Specialist and Data Specialist - Sigma.AI

LabelboxAudioClassification

Support semantic categorisation and intent/entity extraction work for conversational AI and virtual assistant applications through high-volume data operations. Analyze localized dialectal variations, slangs, and idioms to develop stronger training sets for natural language understanding. Evaluate and resolve training dataset inconsistencies while translating linguistic constraints into implementable ML features. • Perform high-volume operations for semantic categorisation, intent classification, and entity extraction • Incorporate East African dialectal and idiomatic variation into training dataset preparation • Identify and fix dataset inconsistencies to improve pipeline readiness • Coordinate with ML engineering teams to map linguistic and syntactic relations into technical features

2021 - 2021

Show more

Education

Work History