Summary
Overview
Work History
Education
Accomplishments
Skills
Certification
Timeline
Generic
Aaron Maupin

Aaron Maupin

Senior Machine Learning Engineer
Augusta,GA

Summary

Machine Learning Engineer with a decade of experience in AI, Machine Learning, and Data Analytics, adept in integrating advanced technologies like Optical Character Recognition (OCR), Generative AI (GenAI), Large Language Models (LLM), Convolutional Neural Networks (CNN), and Deep Neural Networks (DNN) into complex systems. Skilled in areas of Predictive Analysis, Data Manipulation, Mining, and Visualization, with a strong foundation in Business Intelligence. Proficient in training algorithms on diverse cloud platforms, catering to industries such as finance, marketing, advertising, geospatial analysis, and IoT. Expertise in Natural Language Processing and Computer Vision, with a track record of constructing effective data ETL and machine learning pipelines, and implementing end-to-end MLOps solutions. Recognized for delivering seamless, impactful solutions in dynamic and challenging environments.

Overview

10
10
years of professional experience
5
5
years of post-secondary education

Work History

Senior Machine Learning Engineer

Apple
Los Angeles, CA
11.2021 - 12.2023
  • Provided deep learning solutions for several internal use-cases using custom-trained/fine-tuned, BERT, LLM, sentence embedding, cross encoder and collaborative filtering models; First fine-tuned LLM to incorporate human feedback at Apple.
  • Developed and deploy deep learning model to predict tags for blog posts on Apple Education web app using LLM, sentence embedding model, and custom trained/fine-tuned BERT model. (education.apple.com)
  • Crafted and improved Python libraries, tools and dashboards to: develop, train and evaluate models; automate model training, evaluation and deployment; monitor model and data extraction performance; gather human feedback for further training; and make efficient use of cloud resources.
  • Productionalized deep learning models to classify input in public and enterprise facing Apple support apps using custom trained/fine-tuned HuggingFace transformers BERT. First fully automated end-to-end model pipeline at Apple. (1M+ queries per month) (support.apple.com)
  • Freed up fellow team members by implementing tools to automatically extract, transform, load, clean and curate data on cloud infrastructure using S3, Postgres and Oracle DB, replacing existing manual tasks.
  • Served as the Tech lead for implementing NLP for highly complex languages like Finnish, Russian, and Dutch, enabling Siri to understand complex linguistic concepts, such as inflections, in question answering.
  • Scaled Apple's video search backend to support all international languages, thereby consolidating the data source serving Siri, Safari, and Spotlight search.
  • Led a globally distributed 6-person group in evaluating and enhancing Siri's performance in European languages for music and movie playback, resulting in the successful launch of HomePod and Apple TV in the European market.
  • Mentored and onboarded new hires and junior developers in areas of coding standards, source control, Agile methodology, and in-house infrastructure.

Graduate Research Student

University Of Georgia
, GA
08.2020 - 10.2021
  • Expanded automated diagnosis of post-stroke aphasia, using a large dataset of 2635 transcripts from 546 participants, including Broca's, Wernicke's, Anomic, and Conduction aphasia.
  • Employed a language model-based classification method, effectively distinguishing between different aphasia types and healthy controls, demonstrating significant improvements over baseline classifiers.
  • Demonstrated that the type of discourse task influences classification accuracy, with procedural tasks enhancing performance.
  • Utilized deep learning methods, eliminating the need for manual feature engineering, and allowing models to develop their own feature representations.
  • Confirmed that language models became sensitive to known clinical indicators of aphasia, such as closed-class lexical elements in Broca's aphasia and filled pauses in Anomic aphasia.
  • Identified the influence of main elements in discourse tasks as critical for accurate classification in Wernicke's and Conduction aphasia.
  • Showed the potential of language models to uncover multiword features and contribute to understanding language production in aphasia.

Machine Learning Engineer

DataRobot
Los Angeles, CA
07.2019 - 07.2020
  • Significantly improved fraud detection accuracy to 95%.
  • Coordinated with different functional teams to implement models and monitor outcomes.
  • Developed and leveraged innovative features and algorithms to drive down false positives and identify perceived threat across the claims.
  • Utilized traditional statistical analytics, graph theory / network science, ensemble methods and Natural language processing, text analytics, factors analysis / construct development and testing, machine learning feature development and engineering, etc.
  • Employed ensembled methods to increase the accuracy of training model with different Bagging and Boosting method (Xgboost, LGBM, Gradient Boosting, Adaboost).
  • Used cross-validation to test the models with different batches of data to optimize the model and prevent overfitting.
  • Work on feature selection and feature engineering to determine features that have correlation with fraud patterns.
  • Participating in the deployment of machine learning models into production environments, ensuring seamless integration with existing systems.
  • Tuning the performance of ETL processes and data pipelines to ensure efficiency, scalability, and responsiveness.
  • Recommend actionable steps that improve accuracy of the prediction models, policy provisions.
  • Keep up with new fraud trends by performing quantitative and qualitative analysis to identify new fraud patterns.
  • Training and Knowledge Sharing training and knowledge-sharing sessions with team members on new technologies, tools, and methodologies.

Machine Learning Engineer

Microsoft
San Francisco, CA
01.2018 - 06.2019
  • Mined logs and ad corpus, tagging Political and 3PGS terms, and extracted extensive features from multiple data sources using SQL and Python. Implemented data cleaning and mining techniques.
  • Calculated query embeddings and similarities with TwinBERT; trained and fine-tuned NLP models including DNN, BERT, CULR, and GBDT using Jupyter, scikit-learn, Pytorch, Azure ML Studio.
  • Designed sampling strategies and models to address imbalanced train datasets using SMOTE, Cost-sensitive Learning, and Data augmentation.
  • Built and maintained large-scale pipelines for bi-weekly and daily inference on Full Ad Corpus (billions of rows) using Aether, Scope, C#.
  • Conducted A/B Test experiments resulting in a 68% drop in 3PGS violation rates, a 47% drop in Political violation rates, and a 4% reduction in overall pDefect rate.
  • Developed a Hitapp for label collection from human judges and built dashboards in Power BI to monitor policy violation rates and crucial metrics (PClick, filtration count, revenue) at various levels.
  • Collaborated with the tech team for online model implementation on the server using C++.

Machine Learning Engineer Intern

Microsoft
San Francisco, CA
05.2017 - 07.2017
  • Owned the whole life cycle of a machine learning model in production, from data collection and analysis, feature extraction, model training, evaluation, deployment, and refresh and maintenance.
  • Trained the developed models and run evaluation experiments on customers reviews.
  • Performed statistical analysis of results and fine-tuning models.
  • Professional competency in Statistical NLP / Machine Learning, especially Supervised Learning - Document classification, information extraction, and named entity recognition in-context.
  • Worked with Proof of Concepts (POC's) and gap analysis and gathered necessary data for analysis from different sources, prepared data for data exploration using data wrangling.
  • Strong SQL Server and Python programming skills with experience in working with functions.
  • Interacted with data from Hadoop for basic analysis and extraction of data in the infrastructure to provide data summarization.
  • Created visualization tools and dashboards with Tableau, ggplot2 and d3.js.

Machine Learning Engineer Intern

Nucleus Security
Menlo Park, CA
04.2015 - 08.2016
  • Led the data collection, organization, and cleaning of more than 10 hand segmentation datasets. Trained yolov8 COCO hand model for precise bounding box detection and efficient data preprocessing, ensuring optimal storage, achieving a 15% increase in depth and mask accuracy.
  • Fine-tuned a YOLOv8-based hand segmentation model on GPUs by integrating datasets to achieve a balanced dataset.
  • Enhanced model accuracy from an mAP50-95 of 0.75 to 0.91, leading to a 30% improvement in model reliability and efficiency for in-depth ergonomic evaluations.

Undergraduate Student Researcher

University Of Georgia
Athens, GA
02.2014 - 03.2015
  • Research in machine learning and computer vision to predict human affectation from facial expression data and movie clips using OpenCV, Matlab and scikit.
  • Developed distributed python code to facilitate mathematics network and graph theory research.

Education

Master of Science - Computer Science

University of Georgia
Athens, GA
07.2021 - 05.2022

Bachelor of Science - Computer Science

University of Georgia
Athens, GA
09.2013 - 12.2017

Accomplishments

Data Engineering and ETL:

Extensive experience as a data engineer building ETL pipelines.

Proficient in ETL tools in diverse platforms: AWS, Azure, DataBricks, Apache Spark, and Airflow.

Worked with various cloud platforms, emphasizing expertise in AWS and Azure.

Continuous Integration and Deployment:

Utilized CI/CD tools such as AWS and GitLab to ensure seamless integration and deployment of NLP solutions.

Wrote automation processes in Python, leveraging AWS Lambda for efficient and automated workflows.

Employed Docker for deployment on diverse platforms, including Linux, Windows, OSX, and AWS.

Big Data and Cloud Platforms:

Reviewed and implemented data ingestion and initial analysis processes using MongoDB, node.js, and Hadoop.

Designed and deployed cost-effective infrastructure on AWS, optimizing functionality.

Scaled analytics solutions to Big Data using Hadoop, Spark/PySpark, and other relevant tools. Experience with Public Cloud platforms such as Google Cloud, Amazon AWS, and Microsoft Azure.

Machine Learning and Analytics:

Applied machine learning techniques, including regression, classification models, deep learning neural networks.

Designed star schema, Snowflake schema for Data Warehouse, and ODS architecture.

Leveraged advanced statistical procedures for Supervised and Unsupervised problems.

Visualization and Presentation:

Designed visually stunning visualizations using Tableau for effective communication and presentation.

Published and presented dashboards and Storyline on web and desktop platforms.

Database and SQL:

Worked with relational databases, including Teradata and Oracle, demonstrating advanced SQL programming skills.

NLP and Text Processing:

Applied various NLP methods for information extraction, topic modeling, parsing, and relationship extraction.

Developed, deployed, and maintained scalable production NLP models using NLTK, SpaCy, and transformers for automated customer response.

DevOps and Data Platform:

Demonstrated familiarity with DevOps practices and tools specifically tailored for data platform development and deployment.

Proficient in utilizing tools such as Azure DevOps, GitHub, Azure Resource Manager, Azure CLI, PowerShell, and ARM templates for streamlined and efficient development processes.

Skills

Data Science Specialties: Natural Language Processing, Machine Learning, Internet of Things (IoT) analytics, Social Analytics, Predictive Maintenance, Stochastic Analytics

Analytic Skills: Bayesian Analysis, Inference, Models, Regression Analysis, Linear models, Multivariate analysis, Stochastic Gradient Descent, Sampling methods, Forecasting, Segmentation, Clustering, Naïve Bayes Classification, Sentiment Analysis, Predictive Analytics, Econometrics modelling

Analytic Tools: Classification and Regression Trees (CART), H2O, Docker, Support Vector Machine, Random Forest, Gradient Boosting Machine (GBM), TensorFlow, PCA, RNN, Linear and non-Linear Regression, Decision Tree, Support Vector Machine, Stochastic Gradient Boosting, Xgboost, LGBM

Visualization Analytic tools: Power BI, Tableau, R, Python, Matplotlib, Plotly, Seaborn

Analytic Languages and Scripts/ETL tools: R, Python, HiveQL, Spark, Spark MLlib, Spark SQL, Hadoop, Scala, Impala, MapReduce, AWS S3, AWS glue, EMR, AWS redshift, AWS lambda, Azure Data Factory, Azure Data Lake gen 2, Azure Synapse, Apache Airflow, Dags, Data Bricks, Snow Flake

Languages: Java, Python, R, Scala, C/C, JavaScript, SQL, SAS

Python Packages: Numpy, Pandas, Scikit-learn, Tensorflow, Keras, SciPy, Matplotlib, Seaborn, Plotly, NLTK, Scrapy, Gensim, Pytorch

Version Control, CI/CD: GitHub, Git, SVN, Gitlab, Gitbucket, Jenkins, Kubernetes

IDE: Jupyter Notebook, VSCode, Intellij IDEA, Spyder, Eclipse

Data Query: Azure, Google Bigquery, Amazon RedShift, Kinesis, EMR; HDFS, RDBMS, SQL, MongoDB, HBase, Cassandra and NoSQL, data warehouse, data lake and various SQL and NoSQL databases and data warehouses

Deep Learning: Machine Perception, Data Mining, Machine Learning algorithms, Neural Networks, TensorFlow, Keras

Large Language Models: Lang Chain, Open AI

Certification

Neural Networks and Deep Learning

Credential URL: https://www.coursera.org/account/accomplishments/records/K9ZG3Q5ZZMXB

Credential ID: K9ZG3Q5ZZMXB

Introduction to Generative AI

Credential URL: https://coursera.org/share/1653218909455f556a4048e57fc78b60

Credential ID: JH3QSTTZPFHD

Timeline

Senior Machine Learning Engineer

Apple
11.2021 - 12.2023

Master of Science - Computer Science

University of Georgia
07.2021 - 05.2022

Graduate Research Student

University Of Georgia
08.2020 - 10.2021

Machine Learning Engineer

DataRobot
07.2019 - 07.2020

Machine Learning Engineer

Microsoft
01.2018 - 06.2019

Machine Learning Engineer Intern

Microsoft
05.2017 - 07.2017

Machine Learning Engineer Intern

Nucleus Security
04.2015 - 08.2016

Undergraduate Student Researcher

University Of Georgia
02.2014 - 03.2015

Bachelor of Science - Computer Science

University of Georgia
09.2013 - 12.2017

Similar Profiles

CREATE PROFILE
Aaron MaupinSenior Machine Learning Engineer