Keith Thanh - Senior AI Engineer

Keith Thanh

4.9

Select me!

Keith is available for hire

Hire Now

Keith Thanh

4.9

Gender

Male

Main expertise

AI Machine Learning Deep Learning Data Scientists Data Cloud AWS AWS Cloud AWS Lambda Amazon S3 Data Migration Data Visualization

Introduction

Keith Thanh is a highly skilled professional with a versatile skill set, boasting over 12 years of experience in Machine Learning and NLP technologies. He is an experienced leader, with more than 4 years of leading impactful machine learning projects in both the insurance and e-commerce sectors. Keith has notable experience in team building, having recruited, onboarded, and mentored teams through coaching and mentorship. At Google EU, he led the team in the GEO Machine Learning Intelligence project, which focused on employing AI for Google Maps, showcasing his innovative abilities. Keith is adept at managing cross-functional AI teams, with a strong emphasis on communication and collaboration. He holds expertise in cloud services, including AWS, Kubeflow, and Airflow for AI development and deployment. Keith has been recognized for his research excellence with a Ph.D. in Machine Learning, particularly known for his contributions to Semi-Supervised Learning. As a continuous learner, he is certified in SAFe and possesses additional skills in PySpark, data warehousing, and healthcare data analysis.

Project list

Geo - Machine Learning Enhancement - Ireland

Utilized large language models to enhance the accuracy and performance of Google Maps services.

Managed the project pipeline, ensuring high-quality deliverables and timelines were met. Established and maintained clear communication channels with US stakeholders, translating complex technical processes into business outcomes for non-technical audiences. Integrated advanced technologies such as stream-based methods, online learning, and reinforcement learning to drive innovation and maintain industry leadership.

Technologies: Confidentail

Language Model Evaluation and Development - Ireland

Led collaboration and research efforts across functional teams to develop a Proof of Concept.

Evaluated large language models using an API integrated with RAG and Qdrant. Developed a PDF chatbot for the healthcare domain using Groq, OpenAI, and Hugging Face platforms. Enhanced LLMs-RAG by incorporating a knowledge graph.

Technologies: AWS, Kuberflow, Airflow, LLMS, Lamma Index, Langchain, OpenAl Gateway, NetworkX, Neo4j, Streamlit, Evidently

Underwriting AI - Ireland

Led an Underwriting AI project focused on developing and analyzing models to enhance decision-making, profitability, and customer retention.

Led a team of six, providing daily mentorship and coaching to foster professional growth. Led an Underwriting AI project from inception to successful completion, formulating strategies for model development and analysis to boost decision-making, profitability, and customer retention. Managed the project using Agile, Jira, and Scrum methodologies to ensure efficiency, risk management, and effective task allocation. Established strong communication channels with US stakeholders, translating technical processes into business insights for non-technical audiences. Continually integrated cutting-edge innovations such as Stream-based methods, Online Learning, Reinforcement Learning, and technologies like Spark and Airflow to maintain industry leadership.

Technologies: AWS, Kuberflow, Airflow, LLMS, Lamma Index, Langchain, OpenAl Gateway, NetworkX, Neo4j, Streamlit, Evidently

Healthcare Claims System Transformation - Ireland

This project involved transforming the healthcare claims system by transferring data to a new infrastructure, effectively managing claims, and enhancing model operations.

Led a team of 3 Data Scientists in successfully transferring the US healthcare Claim system to the Ireland data lake. Managed hierarchical parsing of the claim system, achieving substantial savings in Global Surgery Package Claims and Outpatient Facility Claims. Spearheaded the construction of base claim models and the development of a Closed Loop pipeline, elevating claim models to MLOps.

Technologies: AWS, Kuberflow, Airflow, LLMS, Lamma Index, Langchain, OpenAl Gateway, NetworkX, Neo4j, Streamlit, Evidently

Customer Data Deduplication and Monitoring System - Ireland

Designed and implemented a deduplication algorithm that enhanced customer data processing efficiency, alongside monitoring and deployment solutions.

Engineered a deduplication algorithm for customer records, significantly improving data processing efficiency. Developed a monitoring pipeline for daily Model & Data Drift using Terraform and AWS SageMaker Step Function, ensuring model accuracy and minimizing errors. Collaborated with cross-functional teams to deploy Machine Learning models on the AWS cloud seamlessly. Conducted comprehensive testing and analysis to guarantee model reliability and accuracy.

Technologies: AWS, Sagemaker, Lambda functions, Evidently

Telecom Fault Diagnostics Project - Ireland

Led a team to develop and deploy a Probabilistic Graphical Model for root-cause analysis in telecom faults, significantly speeding up issue identification and resolution.

Led a team in developing and deploying a Probabilistic Graphical Model for Root-Cause analysis in Telecom Faults, expediting issue identification and resolution. Contributed to algorithms for Intelligent Filtering and Anomaly Detection in Telecom Time Series Signals, enhancing data processing accuracy. Collaborated seamlessly with cross-functional teams, ensuring the successful deployment of the Telecom Fault Diagnostics model. Conducted rigorous testing and analysis to enhance model reliability and accuracy.

Technologies: KL Divergence, Probabilities Graphical Model, Libpgm

Advanced Mobile Data Analysis and Prediction Project - Ireland

This project focuses on analyzing mobile data to identify suspect relationships and applies machine learning for industry applications, such as product repricing and audience forecasting.

Developed a comprehensive framework for analyzing mobile data with suspect relationships, collaborating with law enforcement partners. Led research and development for a scalable machine learning project on product repricing, partnering with industry leader Xsellco Ltd. Coordinated projects on tracking evolving news stories and interpretable classification of human movement videos. Conducted groundbreaking research in predicting and forecasting the total audience size for television shows in collaboration with BRC Ventures Ltd.

Technologies: OpenCV, Timeseries SAXSEQ, Pose Estimation

Project list

Education

Expertise

Certifications