Kieron A.

Key Skills

Software

Other

AWS SageMaker

Top Subject Matter

Text-to-SQL / GenAI evaluation

MLOps for ML experimentation and deployment

Predictive modeling for ROI/lead scoring/user journeys

Top Data Types

Text

Image

Top Task Types

Text Generation

Function Calling

Freelancer Overview

Feature Store; GenAI POC (natural language to SQL). Brings 8+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other and AWS SageMaker. Education includes Master of Science, University of Texas at Austin (2024) and Bachelor of Science, University of Malaya (2019). AI-training focus includes data types such as Text, Computer Code, and Programming and labeling workflows including Text Generation, Function Calling, and Evaluation.

Labeling Experience

Machine Learning Engineer (MLOps work and experimentation enablement)

AWS SageMakerFunction Calling

Developed and operated MLOps/ML pipeline components that enabled repeated model training, experimentation, and evaluation in support of production deployment. Implemented CI/CD test improvements to reduce deployment failures and increase reliability of training and serving workflows. Integrated real-time model serving infrastructure to support ongoing experiment-to-production iteration. • Set up MLOps tooling for training compute, pipelines, feature store, artifact storage, experiments, serving, and monitoring. • Improved CI/CD pipelines by adding tests to reduce deployment failures. • Deployed real-time models with Flask, Docker, AWS ECR, CodeFresh, and Kubernetes. • Coordinated with data engineering to store prediction artifacts in Parquet and Iceberg, cutting storage costs by 70%.

2022 - 2025

Feature Store; GenAI POC (natural language to SQL)

OtherTextText Generation

Built and evaluated a GenAI-style natural language to SQL proof of concept by testing open-source LLMs against GPT for text-to-SQL conversion tasks. Designed the evaluation approach to compare model outputs for translating user text into structured SQL queries. Used these experiments to assess readiness for downstream use cases and model behavior in production-like settings. • Tested open-source LLMs (nsql-llama-2-7B, codellama-7B, SQLcoder2-7B) versus GPT. • Focused on text-to-SQL quality and functional correctness for query generation. • Performed comparative evaluation to select candidate models for further work. • Connected GenAI experimentation to an ML product workflow supporting fraud/credit/transaction enrichment.

2021 - 2025

Data Scientist (modeling and experimentation)

OtherText

Performed data-driven analytics and experimentation work to improve marketing and product outcomes using predictive modeling and hypothesis testing. Applied Bayesian inference to estimate campaign ROI and accelerate experiment cycles, effectively supporting training signal generation for models used in decision-making. Built and evaluated classification and reranking models to predict user outcomes and conversion behavior. • Used Bayesian inference to predict campaign ROI and reduce experiment time from 90 days to under 3 days. • Built an imbalanced classification model for lead deposit prediction based on first-day activity. • Designed dashboards and data models for paid campaigns to monitor performance and inform modeling. • Conducted clickstream-based modeling (Neo4j) for user journey POCs used to drive subsequent training/modeling.

2019 - 2021

Education

Work History