Limuel Patrick G.

Key Skills

Software

Clickworker

CloudFactory

Data Annotation Tech

Dataloop

Labelbox

Mercor

Micro1

Mindrift

OneForma

Remotasks

Telus

Toloka

Top Subject Matter

LLM alignment

RLHF data quality workflows

multilingual NLP localization

Top Data Types

Document

Image

Computer Code Programming

Top Task Types

RLHF

Data Collection

Prompt + Response Writing (SFT)

Fine-tuning

Question Answering

Evaluation/Rating

Computer Programming/Coding

Transcription

Text Summarization

Text Generation

Object Detection

Bounding Box

Polygon

Segmentation

Classification

Entity (NER) Classification

Cuboid

Function Calling

Freelancer Overview

Lead AI Data Specialist | Toptal (Remote – U.S. Client Base). Brings 8+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose, Other, and Apache Spark. Education includes Bachelor of Science, Jomo Kenyatta University of Agriculture and Technology (JKUAT) (2022) and Professional Certificate, Moringa School (2025). AI-training focus includes data types such as Text and labeling workflows including RLHF, Translation, and Localization.

Labeling Experience

Senior MLOps & Infrastructure Architect | Kenya Tech Infrastructure Initiative

Text

Designed and implemented model monitoring and governance systems to track data drift and model decay, enabling automated retraining cycles for high-frequency trading models. Built monitoring suite capabilities using ML lifecycle tooling to ensure that evaluation signals triggered ongoing model updates. Managed operational quality monitoring that supports continuous training and improved model performance. • Tracked data drift and model decay in real time. • Triggered automated retraining cycles to maintain accuracy. • Implemented monitoring with MLflow and DVC. • Supported model governance for production-grade ML systems.

2022 - Present

Lead AI Data Specialist | Toptal (Remote – U.S. Client Base)

Don't discloseTextRLHF

Led data quality workflows for an LLM alignment effort using RLHF to improve model reasoning and reduce hallucinations in the Astra testing project. Focused on curating and validating training/assessment data to improve downstream performance for AI product use in East Africa. Coordinated evaluation and quality loops with emphasis on measurable accuracy gains. • Reduced hallucination rates by 22% through RLHF-driven optimization. • Built data quality workflows for model alignment and testing. • Supported multilingual localization outputs by improving reasoning quality. • Worked with validation loops to ensure the training data met quality targets.

2022 - 2025

Lead Predictive Modeling Specialist | Global Inventory Solutions (Freelance/Toptal)

Don't discloseTextFine-tuning

Built demand forecasting and risk mitigation modeling systems that required iterative dataset preparation and target validation for operational prediction outputs. Engineered ensemble forecasting architectures and validated complex prediction targets such as safety-stock levels under seasonal and disruption scenarios. Applied probabilistic modeling to support operational decisioning with reliability targets. • Developed ensemble model combining LSTM and Gradient Boosted Trees for 50+ hubs. • Defined and validated safety-stock prediction targets under real-world volatility. • Implemented Bayesian stock-out early warning system for preemptive routing. • Achieved 99.2% fulfillment rate through improved risk prediction reliability.

2021 - 2025

Senior Data Scientist & ML Engineer | Upwork (Freelance)

TextPrompt + Response Writing (SFT)

Designed RAG pipelines for generative AI systems that query large internal document collections with high accuracy. Built retrieval-augmented question answering functionality using vector search and orchestration tools to support reliable outputs for fintech compliance workflows. This work required preparing and integrating document corpora and evaluation for response quality. • Built a RAG pipeline using Pinecone and LangChain. • Enabled querying 10,000+ internal documents with 94% accuracy. • Reduced time needed for compliance audits by improving retrieval quality. • Integrated retrieval components to improve generative response trustworthiness.

2020 - 2022

AI Data Engineer | Upwork (Freelance – International)

TextData Collection

Engineered and managed production data pipelines and quality monitoring to support machine learning model readiness for customer support and other NLP use cases. Built ETL workflows for datasets at multi-million record scale and implemented data-drift monitoring to protect model performance over time. Ensured validation and integrity checks so that production-level models could train and infer reliably. • Implemented automated data-drift monitoring and validation scripts. • Managed ETL processes for datasets exceeding 5M+ records. • Reduced data processing latency by 50% using Apache Spark. • Ensured 99.9% data integrity for production machine learning models.

2020 - 2022

Show more

Education

Work History